Multidisciplinary Ophthalmic Imaging Group 139 Clinical

Transcription

ARVO 2015 Annual Meeting Abstracts by Scientific Section/Group - Multidisciplinary Ophthalmic Imaging Group
139 Clinical ocular imaging
Sunday, May 03, 2015 1:30 PM–3:15 PM
Exhibit Hall Poster Session
Program #/Board # Range: 587–614/B0045–B0072
Organizing Section: Multidisciplinary Ophthalmic Imaging Group
Contributing Section(s): Physiology/Pharmacology, Retina
Program Number: 587 Poster Board Number: B0045
Presentation Time: 1:30 PM–3:15 PM
Objective analysis of outer retinal layers imaged by optical
coherence tomography in patients with Stargardt disease
Jason C. Park1, Frederick T. Collison2, Gerald A. Fishman1, 2, Rando
Allikmets3, 4, Jana Zernant3, J Jason McAnany1. 1Ophthalmology,
University of Illinois at Chicago, Chicago, IL; 2The Pangere Center
for Hereditary Retinal Diseases, The Chicago Lighthouse for People
Who Are Blind or Visually Impaired, Chicago, IL; 3Department of
Ophthalmology, Columbia University, New York, NY; 4Department
of Pathology and Cell Biology, Columbia University, New York, NY.
Purpose: To develop and apply an objective, semi-automated
algorithm for analyzing outer retinal layers imaged by spectral
domain optical coherence tomography (SD-OCT) in patients with
Stargardt disease.
Methods: Horizontal macular B-scans were acquired (Optos SDOCT/SLO) from 22 normal controls and 20 genetically confirmed
patients with stage 1 Stargardt disease. The number of hyperreflective, outer-retina bands was quantified for each subject using a
novel, semi-automated algorithm. The algorithm detected the number
of hyper-reflective bands based on longitudinal reflectance profiles
(LRPs) as follows: 1) LRPs were generated by averaging 43 adjacent
A-scans (305 mm in width); 2) The number of bands was quantified
based on the second derivative of the LRP. The present analysis
focused on the 3 outermost retinal bands, currently associated with
the retinal pigment epithelium complex (RPE), the cone outer
segment/RPE interdigitation zone (IZ), and the ellipsoid zone (EZ).
To validate the algorithm, band detection was also determined
manually by four graders.
Results: The semi-automated algorithm and manual analysis
showed excellent agreement. The RPE and EZ bands were detected
throughout the entire B-scan in all controls. The RPE band was
detected throughout the entire B-scan in all Stargardt patients,
whereas the EZ band was detected only outside the central lesion.
The extent of IZ band detection differed throughout the B-scan for the
controls and patients. That is, IZ detection for controls was greatest
in the para- and peri-foveal regions (0.5 to 2.6 mm from the fovea),
with a mean detection over these areas of 82%. IZ band detection
for the controls decreased towards the fovea and near-periphery
(mean detection over these areas of 50% and 55%, respectively). In
patients, the IZ band was generally not present in the fovea, para-, or
peri-fovea due to the central lesion. Outside of the lesion, the IZ band
was detected in 26% of the patients (mean detection across the nearperiphery), which is less than half of the detection in controls.
Conclusions: An objective, semi-automated approach can be used
successfully to quantify the number of hyper-reflective, outer-retina
OCT bands. The IZ band is less frequently detected in Stargardt
patients compared to controls, even outside the central lesion,
suggesting abnormalities in outer segment structure and/or RPE
interdigitation.
Commercial Relationships: Jason C. Park, None; Frederick T.
Collison, None; Gerald A. Fishman, None; Rando Allikmets,
None; Jana Zernant, None; J Jason McAnany, None
Support: NIH grants EY019510 (JM), EY001792 (UIC core grant),
EY021163 (RA), EY019861 (RA), and EY019007 (Columbia
University core grant), Pangere Family Foundation, Foundation
Fighting Blindness, Research to Prevent Blindness
High Resolution Imaging of Hard Exudates Using AO-SLO
Muneo Yamaguchi, Shintaro Nakao, Yoshihiro Kaizu, Yoshiyuki
Kobayashi, Takahito Nakama, Shigeo Yoshida, Yuji Oshima, Tatsuro
Ishibashi. Ophthalmology, Kyushu University, Fukuoka-shi, Japan.
Purpose: Hard exudates are observed in fundus of retinal vascular
diseases including diabetic retinopathy. Previous histological studies
in autopsy specimen have shown these characteristics. However, the
detailed characteristics of hard exudates has not been investigated in
vivo patients with retinal vascular diseases.
Methods: Twenty-one eyes of nineteen patients with hard
exudates were examined. We imaged hard exudates using spectral
domain OCT (SD-OCT) and prototype adaptive optics scanning
light ophthalmoscopy (AO-SLO) systems in all patients. The
characteristics (location, size, morphology, and reflection patter) were
investigated.
Results: AO-SLO showed two morphological types of hard exudates
which could not be distinguished from the fundus examination.
One type is accumulated round particles (28.9±5.4mm) (round type;
major axis: 374±236μm, area: 63091±9970μm2, reflective intensity:
134±24). Another type is an irregular shaped hyperreflective
deposition (irregular type; major axis: 227±133μm, area:
24659±8076μm2, reflective intensity:132±19). The area as well as the
major axis in round type were significantly bigger than of irregular
type, whereas there was no significant difference of the reflective
intensity and the vertical location. Both types could be observed in all
diseases. These types could change into each other.
Conclusions: High resolution imaging revealed two types of hard
exudates. One could consist of phagocytes (round type) and another
could be lipid or proteinaceous materials (irregular type). In vivo
observation of hard exudates using AO-SLO confirmed to the
previous histopathological findings.
Commercial Relationships: Muneo Yamaguchi, None; Shintaro
Nakao, None; Yoshihiro Kaizu, None; Yoshiyuki Kobayashi,
None; Takahito Nakama, None; Shigeo Yoshida, None; Yuji
Oshima, None; Tatsuro Ishibashi, None
Trend-based Progression Analysis (TPA) on Macular Thickness
Map to Detect Glaucoma Progression
Chen LIN, Marco Yu, Christopher K. Leung. Department of
Ophthalmology and Visual Sciences, The Chinese University of
Hong Kong, Hong Kong, Hong Kong.
Purpose: The macula has the highest density of retinal ganglion cells
in the retina, representing a strategic location to monitor glaucoma
progression. We evaluated the performance of TPA on the macular
thickness maps generated by optical coherence tomography (OCT)
for detection of glaucoma progression.
Methods: 33 eyes of 18 glaucoma patients were followed at 4-month
intervals for a mean of 7.1years (range: 6.7-7.4 years) for macular
imaging with a spectral-domain OCT (Cirrus HD-OCT, Carl Zeiss
Meditec). Visual field (VF) was performed in the same visits with
Humphrey Field Analyzer (Carl Zeiss Meditec). The macular
thickness data of serial macular thickness maps (each map contained
50x50 superpixel macular thickness measurements) were exported
for TPA for each eye. VF progression was determined using the
©2015, Copyright by the Association for Research in Vision and Ophthalmology, Inc., all rights reserved. Go to iovs.org to access the version of record. For permission
to reproduce any abstract, contact the ARVO Office at [email protected].
Early Manifest Glaucoma Trial criteria. Progressive RNFL thinning
was examined with Guided Progression Analysis (GPA, Carl Zeiss
Meditec).
Results: A total of 726 OCT macular thickness maps were analyzed
and each eye had an average of 22 maps (range: 20-23 visits) for
TPA. 25 eyes (75.6%) showed progressive macular thinning by TPA,
9 eyes showed progressive RNFL thinning by GPA, and 6 eyes had
VF progression. Among the 9 eyes with RNFL progression, 6 also
had macular thinning detected by TPA and all had TPA changes
evident prior to GPA changes. All eyes with VF progression had
macular thinning detected by TPA. Fig.1 illustrates a glaucomatous
eye with progressive macular thinning and RNFL thinning followed
for ~85 months.
Conclusions: A significant proportion of glaucomatous eyes (75.6%)
showed progressive macular thinning over a mean of 7-year followup. Although it remains to be investigated whether the macular
thinning represents disease-related or age-related change, TPA on the
macular thickness map can provide an informative approach to track
the topology of macular thickness changes in glaucoma patients.
FIG.1 A glaucomatous eye of a POAG patient followed for 84.9
months with progressive macular thinning detected by TPA (pixels
encoded in red) before progressive RNFL thinning detected by GPA
(pixels encoded in red).
Commercial Relationships: Chen LIN, None; Marco Yu, None;
Christopher K. Leung, Carl Zeiss Meditec (C), Carl Zeiss Meditec
(F), Carl Zeiss Meditec (R)
Single static images were captured using a flash power of 300 w/
sec with a wide band excitation and barrier filter (535-585 nm and
605-715 nm, respectively) and a 45 degree field of view (FOV). The
zone between the temporal edge of the optic nerve (ON) and macula
(M), designated as the papillo-macular zone (P), of each image was
chosen and analyzed by assigning a grayscale pixel intensity, ranging
from 0 (black) to 256 (white). A grayscale pixel score was determined
for each image (Figure 1). To reduce background noise, the same
images were then “standardized” using a technique, which adjusted
a square area (1.80 x 1.80mm) overlying the ON to an intensity of 0
(Figure 2). In order to determine the consistency of the scoring, the
standardized images were analyzed twice, blindly on two separate
occasions by one observer. Means, standard deviations (SD), t-tests,
and an intra-class correlation (ICC) were performed.
Results: In total, 96 images were analyzed (from the left eye of 32
subjects, mean age=23 yrs). Two eyes had poor quality FAF images
and were excluded from the analysis. In the non-standardized image
readings, the mean pixel intensity for zone P was 84.64±29.27.
After standardization to reduce background noise, the mean pixel
intensity for zone P was highly comparable between the two readings:
38.80±12.81 on the first read (R1) and 38.78 ± 12.84 on the second
read (R2) (p=0.99). The intra-class correlation between the two
standardized readings was 0.99.
Conclusions: This novel standardized technique applied to FAF
images was strongly reliable and reproducible and can yield scores
that quantify autofluorescence in a preselected zone in the fundus.
Clinical applications will allow quantitative scoring of FAF and
follow-up in clinical therapeutic and research settings. A larger study
testing this quantitative scoring system in different demographic
populations are needed to further characterize the full utility of this
modality.
Pre-standardized fundus autofluorescence photo (FAF).
Reliability of a Novel Quantitative Standardized Measurement
Technique for Fundus Autofluorescence Imaging
Grace Huang, Sumana S. Kommana, Bernard C. Szirth, Albert S.
Khouri. Ophthalmology, NJMS, Bridgewater, NJ.
Purpose: FAF images are currently analyzed descriptively and
qualitatively. The purpose is to assess the reliability of a new
standardized measurement technique designed to quantitatively
analyze fundus autofluorescence (FAF) images.
Methods: FAF photos were captured in eyes of healthy subjects
using a Canon CX-1 Hybrid Mydriatic/Non-mydriatic retinal camera.
Post-standardized fundus autofluorescence photo (FAF).
Commercial Relationships: Grace Huang, None; Sumana S.
Kommana, None; Bernard C. Szirth, None; Albert S. Khouri,
None
Optic nerve head and retinal nerve fiber layer OCT values in
a population–based sample of healthy older adults from urban
areas of Parintins, Brazilian Amazon Region
Nivea Nunes Cavascan1, Sung E. Watanabe1, Marcela Cypel1,
Paula Y. Sacai1, Adriana Berezovsky1, Marcos J. Cohen2, Jacob M.
Cohen3, 2, Sergio Munoz4, Rubens Belfort Jr1, Solange R. Salomao1.
1
Departamento de Oftalmologia e Ciências Visuais, Universidade
Federal de São Paulo, São Paulo, Brazil; 2Instituto de Olhos de
Manaus, Manaus, Brazil; 3Departamento de Cirurgia, Setor de
Oftalmologia, Universidade Federal do Amazonas, Manaus, Brazil;
4
Departamento de Salud Publica, Universidad de La Frontera,
Temuco, Chile.
Purpose: The Brazilian Amazon Region Eye Survey (BARES) is
an ongoing population-based study to determine the prevalence and
causes of visual impairment/ blindness in older adults. Spectraldomain optical coherence tomography (SD–OCT) was performed in a
subset of participants from urban areas to investigate optic nerve head
(ONH) parameters and retinal nerve fiber layer (RNFL) thickness in
healthy eyes.
Methods: Subjects were enumerated through a door–to–door
survey and those with ages 45 years and older were invited for
a comprehensive eye exam. ONH e RNFL measurements were
obtained through dilated pupils using iVue–100 SD–OCT in a subset
of participants with healthy eyes. ONH parameters included cupto-disc area ratio, vertical and horizontal cup-to-disc ratios, disc
and rim areas (mm2) and cup volume (mm3). RNFL thickness (mm)
was measured overall and for inferior, nasal, superior and temporal
quadrants. Data from one eye randomly chosen of each participant
were analyzed. The influence of gender, age, IOP and refractive errors
was also analyzed. Statistical significance was considered as p≤0.05.
Results: An overall sample of 1180 adults was examined and
out of these, 64 (36 females) were eligible for this study. Age
ranged from 45 to 65 years (mean=52±5 years). Considering
ONH parameters, mean area, vertical and horizontal cup-to-disc
ratios were respectively 0.34±0.16, 0.53±0.18 and 0.62±0.21.
Mean disc area was 2.34±0.43mm2, rim area 1.50±0.31mm2, and
cup volume 0.10±0.20mm3. The mean overall RNFL thickness
was 104.14±9.72mm, with thickest RNFL in the inferior quadrant
(132.56±15.62mm), followed by the superior (130.33±15.79mm),
nasal (76.95±10.96mm) and temporal (76.48±10.99mm) ones. There
were no significant differences in ONH and RNFL parameters
for gender, IOP or refractive errors. Older age was significantly
correlated with smaller rim area (β=-0.015, p=0.045), thinner overall
RNFL (β=-0.497, p=0.032) and superior RNFL (β=-0.902, p=0.016).
Conclusions: This study provided SD–OCT normative values of
ONH and RNFL parameters for this cohort of healthy older adults
from urban areas of the Amazon region. It suggests that age should
be considered while interpreting these measurements, which are
important for detection of pathological conditions in this population.
Commercial Relationships: Nivea Nunes Cavascan, None; Sung
E. Watanabe, None; Marcela Cypel, None; Paula Y. Sacai, None;
Adriana Berezovsky, None; Marcos J. Cohen, None; Jacob M.
Cohen, None; Sergio Munoz, None; Rubens Belfort Jr, None;
Solange R. Salomao, None
Support: Conselho Nacional de Desenvolvimento Científico e
Tecnológico – CNPq, Brasília, Brasil, Programa Ciência sem
Fronteiras (Grant # 402120/2012-4 to SRS, SM; Research
Scholarships to SRS and RBJ); Fundação de Amparo à Pesquisa
do Estado de São Paulo, FAPESP, São Paulo, Brasil (Grant
# 2013/16397-7 to SRS); Sight First Program – Lions Club
International Foundation (Grant #1758 to SRS).
Peripapillary Tilting Measurement Using Optical Coherence
Tomography in Healthy Japanese Eyes
Shinichi Usui1, Yasushi Ikuno1, Masahiro Akiba2, Nozomu
Matsubara1, Yuriko Yoshinaga1, Tomoko Asai1, Atsuya Miki1, Kohji
Nishida1. 1Ophthalmology, Osaka Univ Graduate School of Med,
Suita, Japan; 2Topcon corp., Itabashi-ku, Japan.
Purpose: We previously reported differing patterns of visual field
loss resulting from peripapillary tilting between eyes with highly
myopic normal tension glaucoma (NTG) and non-highly myopic
NTG. In this study, we evaluated peripapillary tilting in healthy eyes
and analyzed parametric reproducibility.
Methods: Seventy-nine eyes (43 healthy Japanese subjects; mean
age, 45.2±18.5 years; axial length, 23.84±0.98 mm) were enrolled.
We used high-penetration optical coherence tomography (OCT) for
optic disc imaging. To measure peripapillary tilting, 3.4-, 3.7-, and
4-mm-diameter circular scan images around the optic disc were
extracted from 6x6-mm three-dimensional OCT volume images. The
retinal pigment epithelial (RPE) line was traced, the image divided
into 24 clock hours (Fig 1, superior [S] 1-S12 and inferior [I] 1-I12),
and the average RPE line height calculated in each sector (H1). The
peripapillary tilting index (PTI) (pixels) (difference in RPE height
between one sector and its counterpart) was evaluated (H2). Each
sectorial PTI was shown in the central optic disc as 0 pixel. The
sector with the minimal PTI (PTImin) indicated the tilting direction.
Regarding PTI mapping, the ± degrees between the line with the
PTImin and the horizontal line of the optic disc was referred to as
peripapillary tilting degrees, theta. The intervisit PTI reproducibility
for each diameter of the circular scan image was validated twice on
different days based on OCT images from 20 eyes (10 subjects) and
intraclass correlation coefficients (ICC) determined.
Results: The average PTI reproducibility with the three diameters
was good (ICC, 0.84-0.94) in all sectors. In all eyes, ~86% of the
tilting direction was the same among the diameters. The mean
PTImin was -314±224 in the 3.4-, -332±224 in 3.7-, and -358±259 in
4-mm-diameter images; theta was -0.47±30.2 in the 3.4-, -0.09±30.7
in 3.7-, and +0.4±30.8 in 4-mm-diameter images. About 80.8~82.2%
had tilting in four sectors: S1, 30.3~31.6%; I1, 18.9~26.5%; S2,
12.6~15.1%; and I2, 11.2~15.1% with the three diameters.
Conclusions: PTI reproducibility was good. Each PTI value obtained
using the three circular scan diameters was acceptable. The absolute
PTImin value enlarged with the larger circular scan diameter. Most
healthy Japanese subjects had peripapillary tilting in four sectors.
The PTI will be useful to evaluate the morphology of the optic disc
complex.
Commercial Relationships: Shinichi Usui, None; Yasushi Ikuno,
Bayer (C), Novartis (C), Tomey (F), Topcon (F); Masahiro Akiba,
Topcon corp. (E); Nozomu Matsubara, None; Yuriko Yoshinaga,
None; Tomoko Asai, None; Atsuya Miki, Nidek (C); Kohji
Nishida, Abbott Medical Optics Code (F), Otsuka Pharmaceutical
(F), Shionogi (F)
MACULAR THICKNESS MEASURED BY OPTICAL
COHERENCE TOMOGRAPHY IN HEALTHY OLDER
ADULTS FROM THE CITY OF PARINTINS, BRAZILIAN
AMAZON REGION
Sung E. Watanabe1, Paula Y. Sacai1, Nivea N. Cavascan1, Marcela
Cypel1, Sergio Munoz2, Paulo Henrique A. Morales1, Jacob M.
Cohen3, 4, Adriana Berezovsky1, Rubens Belfort Jr1, Solange R.
Salomao1. 1Ophthalmology, Federal University of Sao Paulo, Sao
Paulo, Brazil; 2Salud Publica, Universidad de La Frontera, Temuco,
Chile; 3Instituto de Olhos de Manaus, Manaus, Brazil; 4Divisão
de Oftalmologia, Depto. de Cirurgia, Universidade Federal do
Amazonas, Manaus, Brazil.
Purpose: To provide normative values of macular thickness in a
subset of participants of a population-based ocular eye survey using
SD-OCT.
Methods: The Brazilian Amazon Region Eye Survey (BARES)
is an ongoing cross-sectional population-based study to determine
prevalence and causes of visual impairment, blindness and ocular
diseases in adults aged 45 years and older in urban and rural areas of
the city of Parintins. Sixty-four eyes of 64 healthy normal subjects
(28 males), aging from 45-65 (mean= 52±5) had a comprehensive
ophthalmic exam. All selected participants had best corrected
visual acuity ≥ 20/32, normal eye exam and no prior ocular surgery.
OCT was performed using standard scanning protocols and retinal
full thickness in Early Treatment Diabetic Retinopathy Study
(ETDRS) subfields was assessed. Retinal thickness was calculated
for 9 subfields from the inner and outer retinal boundaries. The
associations between thickness measurements with age, gender,
refractive errors and schooling level were investigated. Statistical
significance was established at ≤0.05.
Results: The average central foveal thickness was 252.3±19.2mm.
Age was negatively correlated with macular thickness in the
following areas: inner foveal layer (p=0.020); inner nasal (p=0.004)
and inferior (p=0.002) parafovea; outer layer of nasal (p=0.016),
superior (p=0.000) and inferior (p=0.045) parafovea; nasal (p=0.001)
and superior (p=0.026) perifovea. Macular thickness showed to be
larger in males in the following areas: temporal parafoveal total
thickness (p=0.0198), temporal inner parafoveal (p=0.0058) and
nasal outer parafoveal (p=0.033). There was no relationship between
macular thickness and spherical equivalent. The data suggested
a negative correlation between schooling educational level and
decrease in retina thickness for inner (p=0.020) and outer foveal areas
(p=0.000).
Conclusions: The current results provide a normative database of
macular thickness in healthy Brazilian Amazon population using SDOCT. We suggest that age should be considered while interpreting
the retina thickness. These parameters will be useful to characterize
retinal disorders in future studies in this population.
Commercial Relationships: Sung E. Watanabe, None; Paula Y.
Sacai, None; Nivea N. Cavascan, None; Marcela Cypel, None;
Sergio Munoz, None; Paulo Henrique A. Morales, None; Jacob
M. Cohen, None; Adriana Berezovsky, None; Rubens Belfort Jr,
None; Solange R. Salomao, None
Support: Conselho Nacional de Desenvolvimento Científico
e Tecnológico – CNPq, Brasília, Brasil, Programa Ciência
sem Fronteiras (Grant # 402120/2012-4 to SRS, SM and JMF;
Research Scholarships to SRS and RBJ); Fundação de Amparo
à Pesquisa do Estado de São Paulo, FAPESP, São Paulo, Brasil
(Grant # 2013/16397-7 to SRS); Sight First Program – Lions Club
International Foundation (Grant #1758 to SRS).
The Application of Extended Field Imaging to Optical Coherence
Tomography
Akihito Uji, Nagahisa Yoshimura. Ophthalmology, Kyoto Univ
Graduate Sch of Med, Kyoto City, Japan.
Purpose: To report and examine a novel technique that applies
wide-field imaging to existing optical coherence tomography (OCT)
systems. Both spectral domain and swept source OCT imaging
systems were evaluated.
Methods: Ten healthy subjects with normal eyes were recruited
into the study. Horizontal and vertical line scans through the fovea
were acquired from all subjects with conventional OCT (Spectralis®
OCT and DRI OCT-1) and with a novel extended field imaging
(EFI) OCT technique. Various other commercially available OCT
systems (RS3000 [Nidek], Cirrus HD-OCT [Carl Zeiss Meditec],
RTVue-100 [Optovue], and 3D OCT-2000 [Topcon]) were used to
image the eye in one subject for comparison. Differences in scan
length measurements between conventional imaging and EFI OCT
were analyzed. A model eye (developed by Canon, Inc.) with known
dimensions was also imaged using the EFI technique. Scan length
and image distortion in both horizontal and vertical scans were
measured.
Results: Average horizontal and vertical scan length measured 9416
± 410μm and 9437 ± 374μm, respectively, on Spectralis® OCT
scans. When EFI was used with this device, the average horizontal
and vertical scan length significantly increased to 13,610 ± 843
μm and 14,082 ± 701 μm, respectively (both P < 0.0001). Average
horizontal and vertical scan length both measured 12,729 ± 680
μm on DRI OCT-1 images. When EFI was used with this device,
horizontal and vertical scan length increased to 20,745 ± 1738
μm and 21,467 ± 2478 μm, respectively (both P < 0.0001). Actual
model eye dimensions (marked by scale reflections) and scan length
measurements (measured in pixels) were both linear correlated,
indicating that little image distortion occurred during EFI OCT
imaging. In one subject, the EFI technique was successfully used
with all examined OCT systems, and resulted in an increased scan
length in all systems.
Conclusions: The EFI technique enables OCT imaging of a wider
field than is possible with conventional OCT imaging. This technique
is easy to implement and can be applied to a variety of OCT systems.
Optical coherence tomography (OCT) scans through the fovea
obtained with the DRI OCT-1 system. (Top ) Horizontal line scan
through the fovea obtained with conventional OCT. (Bottom)
Horizontal line scan obtained with extended field imaging (EFI) OCT.
Commercial Relationships: Akihito Uji, None; Nagahisa
Yoshimura, Canon (C), Canon (F), Nidek (C), Nidek (F), Topcon
Corporation (F)
Age-related structural changes to the human visual pathway
determined using in vivo clinical imaging techniques
Heather R. Connor1, Sarah L. Hosking2, 3, David F. Abbott4,
5 1
. Optometry, Deakin University, Melbourne, VIC, Australia;
2
Optometry, University of Melbourne, Melbourne, VIC, Australia;
3
Optometry, City University, London, United Kingdom; 4Florey
Institute of Neuroscience and Mental Health, Melbourne, VIC,
Australia; 5Medicine, University of Melbourne, Melbourne, VIC,
Australia.
Purpose: To determine effects of normal ageing on visual pathway
structures in the human brain using in vivo imaging techniques
including 3T Magnetic Resonance Imaging (MRI)
Methods: 83 healthy participants (mean age 58.6 ± 11y, range
40–82y, 37 female) were included. Imaging the retinal nerve fibre
layer (RNFL) and optic nerve head (ONH) was conducted using
Heidelberg HRTII and Stratus OCT instrumentation. 71 participants
(mean age 57.8 ± 11y, range 40-82y, 31 female) completed 3T MRI
imaging (Siemens Trio). T1, T2 and diffusion images were acquired
and retrobulbar optic nerve diameter (ROND), optic radiations
volume (ORV), fractional anisotropy (FA) mean diffusion (MD),
radial diffusion (RD)(MRTrix software) and visual cortex volume
(VCV: SPM8 software) determined. Linear regression analysis was
used to determine effects of ageing for each structure (SPSS20).
Results: RNFL thickness reduced by 4% per decade (pd) in right
eyes (RE) (p < 0.01) and 3.5% in left eyes (LE) (p < 0.02). ONH
measurements showed reduction in Horizontal Integrated Rim
Width of 4.1% pd (p<0.05) RE and 5.9% pd (p<0.01) LE. Vertical
Integrated Rim Area showed 17.4% reduction pd for LE (p < 0.03)
but the 11.4% reduction RE was not statistically significant. HRTII
measurements showed rim volume reduced 10.8% pd RE (p < 0.03)
and 8.8% LE (p < 0.05). Height variation contour reduced 8.1% pd
RE (p < 0.01) and 5.6% LE (p < 0.01). Only RE 15mm behind-theglobe ROND showed significant reduction of 2.4% pd (p < 0.05).
ORV did not show significant change, but FA reduced by 2.2% pd (p
< 0.02) and diffusion measures increased by between 2.1% - 2.9% pd
(p < 0.01). VCV showed ageing effects only in males, with volume
loss of 0.3% pd right (p = 0.028) and 0.4% pd (p = 0.015) left.
Conclusions: The posterior visual pathway appears more robust to
structural changes compared to the anterior pathway, where normal
ageing changes may confound diagnosis of degenerative diseases
such as glaucoma.
Commercial Relationships: Heather R. Connor, None; Sarah L.
Hosking, None; David F. Abbott, None
Support: NHMRC Scholarship
Ellipsoid Zone Mapping and Volumetric Assessment in Normal
and Ocriplasmin-treated Eyes
Justis P. Ehlers, Yuji Ito, Sunil K. Srivastava. Cole Eye InstituteRetina Service, Cleveland Clinic, Cleveland, OH.
Purpose: Ellipsoid zone (EZ) integrity is a key component of visual
outcomes in retinal disease. Transient EZ loss has been described
following ocriplasmin therapy. Objectively measuring the integrity
of EZ remains difficult. The purpose of this study was to evaluate
a novel analysis tool for EZ mapping with en face visualization
and volumetric assessment in both normal eyes and eyes receiving
ocriplasmin.
Methods: An automated EZ mapping tool was developed for
segmenting the EZ with additional retinal layers, providing linear,
area, and volumetric measurements, as well as en face visualization.
An IRB-approved retrospective study was conducted of eyes
receiving ocriplasmin therapy to assess algorithm function with
manual validation. The fellow eye was used as a control eye in cases
without concurrent pathology. Scans were analyzed both pre- and
post-ocriplasmin.
Results: The normal eyes (n=13) were analyzed with EZ mapping.
The mean EZ-RPE volume was 1.13 +/- .18 mm3 and the mean EZRPE central foveal area was 0.20 +/- .03 mm2. The EZ maps revealed
excellent EZ integrity. Mean EZ map thickness was > 20 microns
in 95% of sampled areas. In the ocriplasmin group (n=15), mean
EZ-RPE volume was 1.06 +/- .19 mm3 and the central area was 0.16
+/- .03 mm2 prior to ocriplasmin injection. The EZ maps revealed
good EZ integrity with a mean EZ map thickness of > 20 microns in
92% of sampled areas. One week following ocriplasmin, both EZRPE volume and area were significantly reduced (0.81 +/- .20 mm3
and 0.12 +/- .04 mm2, respectively, p < 0.005). The EZ maps showed
a dramatic reduction EZ integrity with a mean EZ thickness of >
20 microns in only 62% of sampled areas. At 8 week follow-up, EZ
volume and area improved to baseline levels and EZ map integrity
also improved with EZ thickness > 20 microns in 88% of sampled
areas.
Conclusions: Automated EZ mapping provides in-depth visualization
of alterations in retinal architecture and provides a unique opportunity
for assessing longitudinal EZ dynamics. In this study, EZ mapping
demonstrates the significant transient changes that may be noted with
ocriplasmin therapy. This tool may be useful for evaluating additional
retinal diseases and improving our understanding of the status of the
EZ in retinal pathologies.
Ellipsoid zone mapping prior to (top) and 1-week following
ocriplasmin (bottom) with segmentation (A,B), 3D reconstruction
(C,D), En face mapping (E,F).
Commercial Relationships: Justis P. Ehlers, Alcon (C), Bioptigen
(C), Bioptigen (P), Genentech (F), Leica (C), Synergetics (P),
Thrombogenics (C), Thrombogenics (F), Zeiss (C); Yuji Ito, None;
Sunil K. Srivastava, Allergan (F), Bausch and Lomb (C), Bausch
and Lomb (F), Bioptigen (P), Leica (C), Synergetics (P), Zeiss (C)
Support: NIH/NEI K23-EY022947-01A1; Ohio Department of
Development TECH-13-059
Spectral-domain optical coherence tomography characteristics of
eyes with silicone oil tamponade and post silicone oil removal
Alia Durrani1, Ehsan Rahimy2, Nadim Rayess2, Jason Hsu2.
1
Residency Program, Wills Eye Hospital, Philadelphia, PA; 2Retina
Service, Wills Eye Hospital, Philadelphia, PA.
Purpose: To evaluate and characterize the micro-structural retinal
changes present on spectral-domain optical coherence tomography
imaging (SD-OCT) in eyes that have undergone pars plana
vitrectomy and silicone oil (SO) tamponade for various clinical
conditions.
Methods: We conducted a retrospective, consecutive case series
study of patients who underwent pars plana vitrectomy with SO
tamponade for tractional retinal detachment, rhegmatogenous
retinal detachment, or proliferative vitreoretinopathy and had SDOCT performed both during SO tamponade and after SO removal.
Presence of subretinal fluid (SRF), intraretinal cysts (IRC), epiretinal
membrane (ERM), ellipsoid zone and external limiting membrane
disruption (EZD, ELMD), macular hole (MH), hyperreflective
opacities (HO), central foveal thickness (CFT) and choroidal
thickness (CT) were documented. Duration of SO tamponade
was also noted, along with timing of SD-OCTs in relation to SO
tamponade and removal. Lens status was documented along with
incidence of ERM peeling at the time of SO removal.
Results: Six eyes from six patients met all inclusion criteria. Average
time of SO tamponade was 7.32 months or 223.67 days. Average
time SD-OCT was performed after SO tamponade was 126.25 days.
Average time SD-OCT was performed after SO removal was 436.1
days. While under SO tamponade, IRC were noted in 3/6 (50%)
eyes, SRF in 2/6 (33%) eyes, EZD in 5/6 (83.3%) eyes, ELMD in
5/6 (83.3%) eyes, and HO in 3/6 (50%) eyes. Average CMT and CT
while under SO tamponade were 400.16 and 206.33 mm, respectively.
Post SO removal, IRC were noted in 2/6 (33%) eyes, SRF in 1/6
(16.67%) eyes, EZD in 3/6 (50%) eyes, ELMD in 3/6 (50%) eyes,
and HO in 2/6 (33.33%) eyes. Average CMT and CT post SO
removal were 372.33 and 216.5 mm, respectively.
Conclusions: Structural abnormalities of retinal architecture on
SD-OCT in silicone oil filled eyes appear to improve in some eyes
following oil removal.
Commercial Relationships: Alia Durrani, None; Ehsan Rahimy,
None; Nadim Rayess, None; Jason Hsu, Ophthotech (F), Optovue
(C), Santen (F), Xoma (C)
Distal visual pathway changes prior to retinopathy in
streptozotocin-induced diabetes
Swarupa Kancherla1, Ed Wu2, Kevin C. Chan3, 1. 1UPMC Eye Center,
Eye and Ear Institute, Ophthalmology and Visual Science Research
Center, Department of Ophthalmology, University of Pittsburgh,
Pittsburgh, PA; 2Department of Electrical and Electronic Engineering,
The University of Hong Kong, Hong Kong, China; 3Department
of Bioengineering, Swanson School of Engineering, University of
Pittsburgh, Pittsburgh, PA.
Purpose: Diabetic retinopathy is a leading cause of acquired
blindness. Recent studies indicated that visual function is impaired
prior to diabetic retinopathy in diabetic patients. This suggested the
early involvement of the distal visual pathway in the pathogenesis of
diabetes. We assessed the integrity of the retina and visual pathway
in early experimental diabetes using streptozotocin (STZ)-induced
type 1 diabetic rats and in vivo MRI to determine if the distal visual
pathway in the brain is altered before retina in diabetes.
Methods: Ten week old Sprague-Dawley rats (N=10) were
randomized into 2 groups. Five underwent intraperitoneal injection of
STZ at 65 mg/kg in 0.01M citric acid; The other 5 were injected with
citric acid only and acted as control (CTRL). Diffusion tensor MRI
(DTI) was performed 1 month later using a 7 Tesla MRI scanner.
After DTI, chromium (Cr)-enhanced MRI was performed 1 day after
intravitreal Cr(VI) injection into the left eye. Then, MnCl2 solution
was intravitreally injected into the right eye, and manganese (Mn)enhanced MRI was performed 1 day later. DTI parametric values
including fractional anisotropy (FA), axial diffusivity (λ//) and radial
diffusivity (λ) were extracted from prechiasmatic optic nerves. T1weighted signal intensities were measured bilaterally in retina in Crenhanced MRI and subcortical visual nuclei in Mn-enhanced MRI.
DTI parametric values and Cr and Mn enhancement were compared
between STZ and CTRL using Students’ t-tests.
Results: STZ-induced diabetic rats had significantly smaller weight
gain and higher blood glucose level than CTRL rats before MRI
experiments. DTI revealed significant FA decrease and λ increase
in optic nerves of STZ compared to CTRL (p<0.05). There was no
apparent difference between STZ and CTRL in Cr enhancement in
the retina (131%±52% vs 119%±26%, p>0.05). Anterograde Mn
transport as indicated by Mn enhancement in the lateral geniculate
nucleus (41%±15% vs 55%±20%) and in the superior colliculus
(60%±16% vs 62%±21%) was apparently unaffected in STZ
compared to CTRL (p>0.05).
Conclusions: Our results suggest that STZ-induced type 1 diabetes
first leads to structural changes in the distal optic nerve before MRIdetectable changes in retinal integrity and anterograde transport in
retinocollicular and retinogeniculate pathways. Diffusion tensor
imaging can detect the compromise in visual pathway integrity in
early experimental diabetes.
Commercial Relationships: Swarupa Kancherla, None; Ed Wu,
None; Kevin C. Chan, None
Support: NIH Grant P30 EY008098, UL1 TR000005 (Bethesda,
MD); BrightFocus Foundation G2013077 (Clarksburg, Maryland);
Alcon Research Institute Young Investigator Grant (Basel,
Switzerland); Eye and Ear Foundation (Pittsburgh, PA); Research to
Prevent Blindness (New York, NY)
Relationship between distance-weighted peripheral nonperfusion and diabetic macular edema
Min Sagong2, 1, Jano van Hemert3, Charles C. Wykoff4, 5, Rui
Wang5, Srinivas R. Sadda2, 6, David M. Brown4, 5. 1Department of
Ophthalmology, Yeungnam University College of Medicine, Daegu,
Korea (the Republic of); 2Ophthalmology, Doheny Eye Institute, Los
Angeles, CA; 3Optos plc, Dunfermline, United Kingdom; 4Retina
Consultants of Houston, Houston, TX; 5Ophthalmology, Weill
Cornell Medical College, Houston Methodist Hospital, Houston, TX;
6
Ophthalmology, David Geffen School of Medicine, University of
California, Los Angeles, California, Los Angeles, CA.
Purpose: A new method to weight each pixel of non-perfusion
reflecting its respective distance from the fovea was developed in
the hopes of improving the correlation between retinal ischemia
and the development of diabetic macular edema (DME). In this
study, we characterize the relationship between a distance-weighted
ischemic area and the extent of DME using ultra-widefield (UWF)
angiography.
.
Methods: 40 eyes of 40 patients were enrolled in DAVE trial
(NCT01552408) to assess the efficacy and safety of ranibizumab
monotherapy compared with ranibizumab plus UWF fluorescein
angiography (FA) -guided pan retinal photocoagulation for DME with
retinal ischemia. Steered UWF FA images using the Optos 200Tx
were submitted to the Doheny Image Reading Center for analysis.
Images were montaged and transformed to stereographic projected
images for precise quantification using spherical trigonometry. The
ischemic index was computed as ratio of non-perfused to total visible
retina. In addition, each pixel of non-perfused retina was linearly
weighted based on the distance from the fovea and optimized using
simulated annealing to produce a weighted ischemic index, which
gave more importance to ischemia closer to the posterior pole (i.e.
location of edema). Linear and logistic models were constructed to
determine the relationship between the baseline extent of DME and
various covariates including the weighted and unweighted ischemic
indices.
Results: At baseline, there was no significant correlation between
the extent of central macular thickness and macular volume and the
unweighted ischemic index (r = 0.077, P = 0.653 and r = 0.071, P
= 0.677, respectively). However, after linearly weighting each pixel
based on the distance from the fovea, the correlation between the
extent of central macular thickness and macular volume at baseline,
and ischemic index was improved (though not significant) to r =
0.237, P = 0.147 and r = 0.239, P = 0.129, respectively.
Conclusions: In this cohort, the severity of DME was not correlated
with the extent of peripheral non-perfusion. Although the correlation
improved when the ischemic index was linearly weighted to
emphasize more central non-perfusion, it did not achieve statistical
significance. Consideration of other non-linear weighting functions
may strengthen this relationship and should be the target of future
studies.
Commercial Relationships: Min Sagong, None; Jano van Hemert,
Optos PLC (E), Optos PLC (P); Charles C. Wykoff, Alcon (C),
Alcon (F), Allergan (F), Allergan (F), Genentech (C), Genentech
(F), Regeneron (C), Regeneron (F); Rui Wang, None; Srinivas R.
Sadda, Alcon (C), Allergan (C), Allergan (F), Allergan (R), Carl
Zeiss Meditec (C), Carl Zeiss Meditec (F), Carl Zeiss Meditec (R),
Genentech (C), Genentech (F), Norvatis (C), Optos (C), Optos
(F), Optos (R), Roche (C); David M. Brown, Abbott (F), Acuceta
(F), Alimera (F), Allergan (C), Ampio (F), Bayer HealthCare (C),
Genentech (C), Genzyme (C), GSK (F), Heidelberg Engineering (C),
Norvatis (C), Notal Vision (C), Opthotech (F), Pfizer (F), pSivida (F),
QLT (C), Quark (F), Regeneron (C), Santen (F), Thrombogenetics
(C)
Comparison Of Non-Mydriatic Ultra-Widefield Fundus
Photography To Standard Early Treatment of Diabetic
Retinopathy Study 7- Fields In Diabetic Retinopathy Screening
James D. Kim1, Clinton Warren1, Shachar Tauber2, Mara R.
Goldberg1, Judy E. Kim1. 1Department of Ophthalmology,
Medical College of Wisconsin, MIlwaukee, WI; 2Department of
Ophthalmology, Mercy Hospital, Springfield, MO.
Purpose: To assess the effectiveness of non-mydriatic ultrawidefield fundus photography (UWFFP) in detecting peripheral
vascular lesions and to compare it to the Early Treatment of Diabetic
Retinopathy Study (ETDRS) 7-fields for screening of diabetic
retinopathy.
Methods: A retrospective review of Optos® camera ultra-widefield
fundus photography obtained during diabetic retinopathy screening.
Two independent reviewers screened each photograph for quality
based on preset paramaters. Then microaneurysms (MA), intraretinal
hemorrhages (IRH), venous beading (VB), intraretinal microvascular
abnormalities (IRMA), cotton wool spots (CWS), hard exudates
(HE), neovascularization on the disc (NVD), neovascularization
elsewhere in the retina (NVE), preretinal hemorrhage (PRH), vitreous
hemorrhage (VH) were counted within and outside ETDRS 7-field
areas and divided into four quadrants. Statistical analyses were
performed to compare retinal abnormalities in respective regions.
Results: 193 eyes from 100 patients were included in this study.
Significant number of outer quadrants (N = 433) had greater
than 50% of its area covered by obscurations and artifacts when
compared to inner quadrants (N = 76, p = 10-64). Full ETDRS 7-fields
without significant artifacts were captured in 149 eyes (77.8%).
Most commonly affected outer quadrants were inferonasal and
inferotemporal quadrants (89% and 64% respectively). MA and IRH
were significantly more frequent within EDTRS 7-field quadrants
than outer quadrants (MA Nin vs. Nout = 116 vs. 20, p = 10-102 and IRH
Nin vs. Nout = 115 vs. 58, p = 10-14 respectively). VB, CWS, and HE
were exclusively present within ETDRS 7-field quadrants. One case
of NVD was seen. No NVE, PRH, or VH were identified. Incidental
findings included macular and peripheral drusen (N = 44), central
retinal vein occlusion (N = 3), and choroidal rupture (N = 1).
Conclusions: Optos non-mydriatic UWFFP is a potentially useful
method to screen for diabetic retinopathy since it allows imaging
of the peripheral retina and ETDRS 7-fields even without dilation.
While the peripheral retina outside of ETDRS 7- field are frequently
obscured by artifacts, especially in inferior quadrants, the majority
of pathology was within the ETDRS 7 fields. Additional peripheral
pathologies were found outside the ETDRS 7-fields that otherwise
would have been missed.
Commercial Relationships: James D. Kim, None; Clinton
Warren, None; Shachar Tauber, None; Mara R. Goldberg, None;
Judy E. Kim, None
Support: Research to Prevent Blindness
Interocular Symmetry in Spatial Distribution of Diabetic Retinal
Lesions
Sung Yong Park1, Da Ye Choi2, Mingui Kong2, Don-Il Ham2.
1
Ophthalmology, Soonchunhyang Univ. Cheonan Hospital, Cheonansi, Korea (the Republic of); 2Department of Ophthalmology, Samsung
Medical Center, Sungkyunkwan University School of Medicine,
Seoul, Korea (the Democratic People’s Republic of).
Purpose: To assess the interocular symmetry in spatial distribution of
diabetic retinal lesions using ultra-wide-field imaging.
Methods: Imaging data of patients who had diabetic retinopathy
and underwent ultra-wide-field angiography (UWFA) were
retrospectively investigated. Fundus in each UWFA image was
divided into 2 or 4 subfields by a grid with horizontal and vertical
straight lines centered at the fovea. Fundus was also divided into
central and peripheral subfields by a rectangle grid. UWFA images
taken just after completion of arteriovenous phase were evaluated
for microaneurysm (Ma), presumed intraretinal microvascualr
abnormalities (IRMA), and non-perfusion area (NPA) on each
subfield. Each subfield was scored in a scale of 0-4 for each retinal
lesion, based on the detected number of corresponding lesion.
Interocular agreements of score in each subfield and interocular
agreements of score difference between different subfields were
analyzed using a Kappa index.
Results: Two hundred eighty eight eyes of 144 patients were
evaluated; 244 eyes (84.7%) had non-proliferative diabetic
retinopathy (NPDR), and 42 eyes (14.6%) had proliferative diabetic
retinopathy. Significant interocular agreement was found in the score
of Ma (Superior/Inferior: 0.67/0.74, Nasal/Temporal: 0.87/0.89,
Central/Peripheral: 0.89/0.80), presumed IRMA (Superior/Inferior:
0.82/0.87, Nasal/Temporal:0.91/0.76, Central/Peripheral: 0.93.0.89),
and NPA(Superior/Inferior: 0.86.0.65, Nasal/Temporal: 0.87/0.81,
Central/Peripheral: 0.97/0.94). High interocular agreement was also
found for the score difference between two vertical half subfields,
two horizontal half subfields, and central versus peripheral subfields (
P<0.05). A kappa index for intragrader and intergrader agreement was
between 0.60 and 0.93.
Conclusions: There was significant interocular symmetry in the
spatial distribution of diabetic retinal lesions. This finding can be
useful for the screening examination and longitudinal evaluation of
diabetic retinopathy.
Commercial Relationships: Sung Yong Park, None; Da Ye Choi,
None; Mingui Kong, None; Don-Il Ham, None
Ultra-Structural SDOCT changes in patients with DMI and
preclinical diseases
Nicole Mendez, Saysha Blazier, Bernard C. Szirth, Albert S. Khouri.
Ophthalmology and Visual Science, Rutgers New Jersey Medical
School, Newark, NJ.
Purpose: To analyze spectral domain optical coherence tomography
(SDOCT) findings in patients with Type 1 Diabetes (DMI) with
normal fundus examination.
Methods: 109 subjects were screened (mean age 14.32, range 5-30
yrs) during The International Conference for Children with DMI in
Orlando, FL. The mean duration of DMI was 7.82 yr (range 0.4 to 25
yr) and the mean HbA1C for these individuals was 7.9%. Of these,
56% were females, 79% were Caucasians and the visual acuity was
20/20 for 57% of right eye and 61% of left eye. A comprehensive
screening was performed including non-mydriatic fundus imaging
(Canon, CR2 Plus-AF with EOS-60D) and SD-OCT (Optovue,
iVue). OCT scans were acquired showing macular thickness (MT)
and thickness of the parafoveal regions (2-4mm from fovea) and
the perifoveal regions (4-6mm from fovea). The subjects were
divided into three age groups: 6-8, 9-14 and 15-30 yr. Associations
of macular, paramacular and perimacular thickness were analyzed as
a function of HbA1C and Body Mass Index (BMI) by simple linear
regressions.
Results: SDOCT measured ultrastructural discrepancies in the
macula, paramacula and perimacula regions. Linear regression
analysis of thickness of macula OD, paramacula OD, paramacula
OS, perimacula OD with BMI were statistically significant (p
<.05). HbA1C >7.5% was also found statistically significant in the
macula OS, paramacula OD, paramacula OS, perimacula OD and
perimacula OS. (Table 1) A general trend of thickening of the macula,
paramacula and perimacula was observed with increasing HbA1C
(>7.5%) and increasing BMI; however, no statistical significance was
found. (Table 2)
Conclusions: SDOCT with retinal imaging was feasible in young
individuals and revealed ultrastructural macular and perimacular
changes in subjects with DMI prior to manifested clinical disease. A
reference data base for comparison will further our understanding of
these changes and warrants a larger study that will include SDOCT in
patients with DMI.
Significant Variables
Thickening observed with increasing HbA1C (>7.5%) and increasing
BMI.
Commercial Relationships: Nicole Mendez, None; Saysha Blazier,
None; Bernard C. Szirth, None; Albert S. Khouri, None
Choroidal, retinal and RPE thickness in diabetic retinopathy
measured with widefield MHz-OCT over 60° field of view
Jan Philip Kolb1, 2, Thomas Klein2, Kathrin J. Mohler2, Wolfgang
Wieser2, Lukas Reznicek3, Marcus Kernt4, Anselm Kampik4, Aljoscha
S. Neubauer4, Robert Huber1. 1Institut fuer Biomedizinische Optik,
Luebeck, Germany; 2Physics, Universität München, Munich,
Germany; 3Ophthalmology, Technische Universität München,
Munich, Germany; 4Ophthalmology, Universität München, Munich,
Germany.
Purpose: To investigate changes in thickness of the choroid, RPE
and remaining retinal layers in patients diagnosed with diabetic
retinopathy over a field of view of 60° (~18mm).
Methods: 15 consecutive patients diagnosed with diabetic
retinopathy (aged 57.4±11.1 years, 41 - 81 years, 3 female and 12
male) were imaged with a wide-field MHz-OCT. This is a custom
swept-source OCT device based on a Fourier-domain mode-locked
laser source with 1050nm central wavelength and a depth scan
rate of 1.68MHz, which is about 16 times faster than the latest
commercial OCT devices. Therefore, a densely sampled scan with
2088x1024 A-scans over 60° took only 1.82s total acquisition time.
In post processing, the retina was divided into 25 sectors: five circles
centered at the macula with diameters of 1mm, 3mm, 6mm, 12mm
and 18mm. The second and third one were divided into 4 sectors,
the outer two into 8 sectors. Right eyes were mirrored such that the
optical nerve head was always in the same sector. A trained observer
manually determined the thickness of choroid, RPE and remaining
retinal layers at 15 points in each sector. Mean and standard deviation
were computed for each sector and layer.
Results: In most patients, segmentation of the top two layers was
possible in sectors that were not affected by shadowing, which is also
known from other widefield devices. Visibility of the choroid sclera
junction was strongly affected by retinal thickness, pigmentation,
eye-blinks and saccades. Among all 15 patients, the choroid could be
segmented in 5, the RPE in 14 and the remaining layers in 14 cases
for at least 20 of the 25 sectors (80%). Results for one patient with
the sector overlay are shown in Figure 1. Mean thicknesses of the
rings for all patients starting from the center are 249, 248, 229, 208,
202 mm for the choroid. 38, 38, 36, 35, 34 mm for the RPE and 291,
346, 314, 261, 252 mm for the remaining layers.
Conclusions: Layer thicknesses of the posterior eye of diabetic
patients over a large area could be measured with OCT for the
first time. Based on the mean values, we could not identify any
characteristics specific for diabetes regarding layer thicknesses
distribution.
a) shows en face projection of one OCT data set with segmentation
sector overlay. b)-d) are the segmentation results of this dataset. Error
bars show standard deviation. Sector 21 could not be segmented due
to strong shading.
Commercial Relationships: Jan Philip Kolb, None; Thomas
Klein, None; Kathrin J. Mohler, None; Wolfgang Wieser, None;
Lukas Reznicek, None; Marcus Kernt, None; Anselm Kampik,
None; Aljoscha S. Neubauer, None; Robert Huber, None
Support: FP7 ERC StG, contract no. 259158; DFG HU1006/3-1;
Gesellschaft der Freunde und Förderer der Augenklinik der LMU
München e. V.
Clinical Trial: DRKS00005173
Assessing endothelial function in type-2-diabetes using a novel
method of ECG-gated retinal artery flicker provocation
Anchal Lal1, Paul Mitchell2, Gerald Liew2, Aravinda Thiagalingam3.
1
Ophthalmology/Cardiology, University of Sydney, Sydney, NSW,
Australia; 2Ophthalmology, University of Sydney, Sydney, NSW,
Australia; 3Cardiology, University of Sydney, Sydney, NSW,
Australia.
Purpose: Endothelial dysfunction is strongly associated with type
2 diabetes and cardiovascular disease. Digital reactive hyperemia
(Endopat®) following brachial ischaemia-induced hyperemia has
traditionally been used to examine endothelial function, however,
it only measures large artery endothelial function. Flicker lightinduced retinal vasodilation has been shown to reflect endothelial
function in the retinal circulation while previous studies have shown
that reduction in retinal vasodilation after flicker light stimulation is
independently associated with diabetes. We aimed to determine if a
novel method of assessing endothelial function, ECG-gated retinal
artery flicker provocation, is associated in subjects with and without
type-2 diabetes mellitus (T2DM), and we confirmed the presence of
endothelial dysfunction with digital reactive hyperemia.
Methods: Thirty non-diabetic age-sex matched control subjects and
30 patients from the outpatient diabetes clinic in Westmead Hospital,
Sydney, Australia were randomly selected to participate in this study.
ECG-gated retinal fundus photographs were taken at 10-second
intervals for a period of 1 minute pre-flicker, 2 minutes with flicker
light-stimulation, and 1 minute post-flicker. A semi-automated
software analysis program, developed in-house, was used to measure
retinal vessel caliber in ECG-gated photographs. Reactive hyperemic
index (RHI) was also examined (EndoPAT 2000). Subjects with
proliferative diabetic retinopathy and cardiovascular complications
were excluded. The independent t-test was used to assess statistical
significance.
Results: The mean age of participants and the duration of diabetes in
subjects with T2DM was 36.9±14.2 and 12.1± 8.8 years respectively.
The majority of subjects were female (62%) and all subjects with
diabetes had minimal to no diabetic retinopathy. Mean arterial
vasodilation in response to flicker was significantly impaired in
subjects with T2DM compared with controls (1.76 ± 0.70 vs. 3.16
± 1.31%, p<0.001). RHI was also significantly reduced in patients
with diabetes compared with controls (1.68± 0.64 vs. 1.98 ± 0.53,
p=0.002).
Conclusions: This ECG-gated retinal imaging system demonstrated
an impairment of retinal artery vasodilation during flicker
provocation in T2DM. This system may represent a useful, noninvasive method for the assessment of endothelial function in patients
with diabetes.
Commercial Relationships: Anchal Lal, None; Paul Mitchell,
None; Gerald Liew, None; Aravinda Thiagalingam, None
Near-infrared reflectance imaging in eyes with acute zonal occult
outer retinopathy
Shinji Ueno, Kenichi Kawano, Yasuki Ito, Ayami Nakanishi,
Masatoshi Nagaya, Hiroko Terasaki. Ophthalmology, Nagoya Univ
School of Med, Nagoya, Japan.
Purpose: To investigate the confocal scanning laser ophthalmoscopic
(SLO) images obtained with near infrared (IR) light in eyes with
acute zonal occult outer retinopathy (AZOOR).
Methods: Observational case series. The medical records of 12 eyes
of 10 patients with AZOOR were reviewed. SLO images obtained
from the AZOOR eyes were compared to images obtained by
spectral-domain optical coherence tomography (SD-OCT), by fundus
autofluorescence (FAF), and by an adaptive optics (AO) fundus
camera.
Results: In 8 of 12 eyes, abnormal hyporeflective areas were
detected in the IR images, and the other 4 eyes did not have specific
abnormalities in the IR images. The boundaries of the abnormal
hyporeflective areas corresponded with the border of the irregularity
of photoreceptor inner segment ellipsoid (ISe) band in the SD-OCT
images. The cone mosaics of the AO fundus image was disrupted in
the abnormal hyporeflective area of the IR image. However, the areas
of FAF abnormalities did not coincide with the hyporeflective areas
in the IR images.
Conclusions: The presence of hyporeflective areas in the IR images
of AZOOR patients suggests impairment of the photoreceptors area.
The IR images would be useful to evaluate eyes with AZOOR.
Commercial Relationships: Shinji Ueno, None; Kenichi Kawano,
None; Yasuki Ito, None; Ayami Nakanishi, None; Masatoshi
Nagaya, None; Hiroko Terasaki, None
Multimodal imaging in Rubella Retinopathy
Danuta M. Bukowska1, 2, Sue Ling Wan1, 3, David A. Mackey1, 2, Fred
K. Chen1, 2. 1Lions Eye Institute, Nedlands, WA, Australia; 2Centre for
Ophthalmology and Visual Science, University of Western Australia,
Perth, WA, Australia; 3Royal Perth Hospital, Perth, WA, Australia.
Purpose: To demonstrate multimodal imaging features of rubella
retinopathy.
Methods: Five patients (age 8 to 49 years) with salt-and-pepper
retinopathy and varying degree of hearing loss underwent complete
ophthalmic examination and multimodal imaging including wide-
field scanning laser ophthalmoscopy, fundus autofluorescence and
spectral domain optical coherence tomography (OCT) (Spectralis,
Heidelberg). Two patients also underwent electrophysiology and 10-2
MAIA microperimetry. Adaptive optics (AO) imaging (rtx1, Imagine
Eye) was performed in one patient to document cone mosaic.
Results: Mean visual acuity was 81 and 79 letters in R and L
eyes. drusen-like structures were present in the macular region in
all eyes. Some of these corresponded to the hyper autofluorescent
spots. Speckled hypo autofluorescent regions were seen within
the posterior pole extending to the periphery to variable extent.
OCT showed subretinal and sub-RPE lesions with diffuse stippled
discontinuity within the ellipsoid band in the macular region. Average
macular retinal sensitivity was 28.7 dB in 4 eyes that underwent
microperimetry. Electrophysiology showed mild generalised cone
and rod dysfunction. Adaptive optics imaging showed reduced cone
density in the perifoveal region .
Conclusions: Hyper autofluorescent spots and diffuse speckled
hypo autofluorescence are common features in rubella retinopathy.
Multimodal imaging may aid clinical diagnosis of this rare condition.
Commercial Relationships: Danuta M. Bukowska, None; Sue
Ling Wan, None; David A. Mackey, None; Fred K. Chen, None
Comparing the Extent of Peripheral Non-perfusion to Central
Non-perfusion in Eyes Affected by Retinal Vascular Disease
Razek Georges Coussa, Cyril Archambault, Mikel Mikhail, David
Elliott Lederer. Opthalmology, McGill University, Montreal, QC,
Canada.
Purpose: Limited data exists comparing the central to peripheral
retinal ischemia in retinal vascular disease. In this study, we
quantitatively compare the extent of non-perfusion in the central 45
degrees to the remaining 200 degrees of angiographically visible
retina in Central retinal vein occlusion (CRVO), Branch retinal vein
occlusion (BRVO), Hemiretinal vein occlusion (HRVO) and Diabetic
retinopathy (DR).
Methods: 95 eyes presenting with retinal vascular pathology to a
single retina specialist between August 2008 and August 2014 were
randomly selected in a retrospective manner. All patients underwent
wide-field retinal imaging, including fundus photography and
fluorescein angiography (FA) using the Optos 200 Tx system (Optos
plc, Scotland, UK). ImageJ 1.43 software was used to calculate the
central areas of non-perfusion (defined as central 45 degrees) and
peripheral areas of non-perfusion (defined as the area beyond 45
degrees). Non-perfusion was defined as dull/dark sections of the FA
occurring in the mid-phase of the test and not attributable to blockage
from blood and/or pigment and/or hard exudate. T-test and linear
regression were used to analyze the data.
Results: The sample included 18 eyes with CRVO, BRVO or
HRVO and 64 eyes with DR. 3 eyes were excluded due to previous
panretinal photocoagualation and 10 due to poor visualization. The
extent of central non-perfusion averaged 9%±11% (range: 0-46%) for
all pathologies combined. Similarly, the average extent of peripheral
non-perfusion averaged 9%±12% (range: 0-51%). These results
demonstrated no significant statistical differences between the percent
central and peripheral non-perfusion areas for both groups (p=0.52
and 0.53, respectively). Linear regression between peripheral and
central non-perfusion areas were statistically significant (p<0.05) for
both groups.
Conclusions: Our data demonstrate a linear relationship between
peripheral and central areas of non-perfusion. Nevertheless, viewing
in isolation misses nearly 42% of the total non-perfusion areas. This
data is critical not only to guide therapeutic options but will also
serve as a gauge for the clinical applicability of medical practices not
using wide field retinal imaging.
Commercial Relationships: Razek Georges Coussa, None; Cyril
Archambault, None; Mikel Mikhail, None; David Elliott Lederer,
Alcon Inc (C), Bayer Inc (C), Novartis Pharmaceuticals Inc (C)
Lens fluorescence and retinal vessel diameter in the Inter99 Eye
Study
Dragana Drobnjak1, Nina Veiby1, Torben Jørgensen2, Line Kessel3,
Michael Larsen3. 1Ophthalmology, Oslo Univ Hosp/Ullevaal, Oslo,
Norway; 2Research Center for Prevention and Health, Glostrup,
Copenhagen, Denmark; 3Department of Ophthalmology, Glostrup
Hospital, Copenhagen, Denmark.
Purpose: We know from previous studies that the risk of ischemic
heart disease is strongly related to lens fluorophore accumulation
and that the relationship is attributable to tobacco smoking and
dysglycemia. Retinal vessel diameter changes are also associated
with smoking, dysglycemia and ischemic heart disease. The
purpose of this study was to examine the relationship between lens
fluorescence and retinal vessel diameter
Methods: The Inter99 study comprised an age- and sex-stratified
sample of 13,016 participants residing in 11 suburban municipalities
of the south-western part of Copenhagen County. Of 6784 subjects
aged 30-60 years who volunteered to participate in the main study,
a subgroup of 970 subjects participated in the eye study. Retinal
blood vessels were measured using a semi-automatic computer
program. All study subjects not having good and clear images (n=62)
were excluded. The analysis comprised a group of 908 subjects.
We investigated the relation between fundus photographic retinal
vessel diameters and lens fluorescence with adjustment for age,
sex, and arterial blood pressure. Vessel diameters were expressed as
central retinal artery equivalent diameter (CRAE) and central retinal
vein equivalent diameter (CRVE). Lens aging was quantified by
lens fluorometry. The degree of cortical lens opacities was graded
according to the LOCS III scale
Results: CRAE increased with increasing lens fluorescence by 0.008
mm/ng f-eq/ml (CI95%: -0.001 to 0.016, p=0.081) after adjusting
for age, sex, and systolic blood pressure. CRVE increased with
increasing lens fluorescence by 0.013 mm/ng f-eq/ml (CI95%: 0.001
to 0.025, p=0.037) after adjusting for age, sex, and systolic blood
pressure
Conclusions: In the Inter99 Eye study, wider vein diameters were
significantly associated with lens fluorescence. The relationship may
be attributable to the metabolic products affecting both the aging
process of the lens and the retinal vessel diameters
Commercial Relationships: Dragana Drobnjak, None; Nina
Veiby, None; Torben Jørgensen, None; Line Kessel, None; Michael
Larsen, None
Choroidal thickness findings in two siblings with high hyperopia
by Swept Source-OCT
Hiroyuki Kaneko, Ari Shinojima, Ryusaburo Mori, Akiyuki
Kawamura, Mitsuko Yuzawa. Ophthalmology, Nihon University,
Tokyo, Japan.
Purpose: To evaluate the choroidal thickness at 5 sites in two
siblings with high hyperopia using Swept Source-OCT (SS-OCT).
Methods: Two siblings (4 eyes) with high hyperopia. Case 1: a
51-year-old woman with high hyperopia (Right: +20.5 Dioptors (D),
Left: +19.5 D). The axial lengths were 15.6 mm in the right eye, 15.6
mm in the left.
Case 2: a 55-year-old man with high hyperopia (Right: +22.5 D, Left:
+22.8 D). The axial lengths were 14.8 mm in the right eye, 14.7 mm
in the left.
Choroidal thickness was measured at 5 sites in each eye using SSOCT; subfoveal, nasal, temporal, superior and inferior (the 4 sites
other than subfoveal were measured 3,000 mm from the fovea).
Results: The mean choroidal thickness was 355.8 ± 63.6 μm at the
subfoveal, 466.3 ± 85.1 μm at the nasal, 274.8 ± 77.2 μm at the
temporal, 396.8 ± 54.6 μm at the superior, and 480.8 ± 66.8 μm at the
inferior site (mean ± SD).
Conclusions: The maximum choroidal thickness was at the inferior
site. The choroid was thin at the nasal, superior, subfoveal and
temporal sites, in that order.
Commercial Relationships: Hiroyuki Kaneko, None; Ari
Shinojima, None; Ryusaburo Mori, None; Akiyuki Kawamura,
None; Mitsuko Yuzawa, None
Location of the tessellation in young Japanese healthy eyes
Takehiro Yamashita, Naoya Yoshihara, Yuya Kii, Minoru Tanaka,
Kumiko Nakao, Taiji Sakamoto. Dept of Ophthalmology, Kagoshima
University, Kagoshima-Shi, Japan.
Purpose: There are several different types of staphyroma in highly
myopic eyes. Tessellation is one of the characteristic myopic changes
in pigmented fundi and some tessellation fundi may progress into the
staphyroma. Therefore we hypothesized there are several types of
tessellation in young healthy eyes. The purpose of this study was to
investigate the location of the tessellation in young healthy eyes and
to compare the myopic changes among the type of tessellation.
Methods: A prospective observational cross-sectional study
comprised 126 right eyes of 126 healthy young Japanese participants.
All participants (mean age 26.0 ± 4.1) underwent comprehensive
ophthalmologic examination, including measurement of axial length
and color fundus photography. Optic disc tilt was quantified using
sine curve approach of the Topcon 3D OCT-1000 Mark II RNFL
3.4 mm circle scan images. We classified ocular fundus into the
following seven categories, such as no tessellation (NO), temporal
(T), infra-temporal (IT), inferior (I), nasal (N), peripapillary (PP),
whole (W), unclassified (U) group, based on the location of the
obvious tessellation. The Steel-Dwass multiple comparison test was
used to analyze the significant difference of axial length, optic disc
plus conus area, optic disc tilt among the groups.
Results: Forty-four eyes were categorized as NO group, 12 eyes as T
group, 21 eyes as IT group, 9 eyes as I group, 8 eyes as N group, 15
eyes as PP group, 11 eyes as W group, and 6 eyes as U group. In the
multiple comparison analysis, for axial length, there was significantly
difference between NO and either IT or W group. For optic disc plus
conus area, there was significantly difference between NO and I, IT
or PP group. For optic disc tilt, there was significantly difference
between NO and IT group (p<0.05).
Conclusions: There was an individual variation in the location of the
tessellation in young healthy eyes. Comparing the NO group, the IT
group had longer axial length, larger optic disc plus conus area, and
larger optic disc tilt.
Commercial Relationships: Takehiro Yamashita, None; Naoya
Yoshihara, None; Yuya Kii, None; Minoru Tanaka, None; Kumiko
Nakao, None; Taiji Sakamoto, None
Clinical Trial: UMIN000006040
Uncertainty in the diagnosis of Pre-plus Disease in Retinopathy
of Prematurity
Allison R. Loh1, Michael Ryan1, Katherine Abrahams1, Esra AtaerCansizoglu4, Robison V. Chan2, Audina M. Berrocal3, Jayashree
Kalpathy-Cramer5, Veronica Bolon-Canedo6, Deniz Erdogmus4,
Michael F. Chiang1. 1Ophthalmology, Oregon Health and Science
University, Portland, OR; 2Ophthalmology, Weill Cornell Medical
College, New York, NY; 3Ophthalmology, Bascom Palmer Eye
Institute, Miami, FL; 4Electrical and Computer Engineering,
Northeastern University, Boston, MA; 5Radiology, Harvard Medical
School, Charleston, MA; 6Informatics, University of A Coruña, A
Coruña, Spain.
Purpose: To measure the impact of pre-plus categorization by
examining the diagnostic accuracy of experts as well as calculating
quantitative tortuosity values using novel computer analysis.
Methods: Wide-angle retinal images captured from 73 infants during
ROP-screening were interpreted by 2 experts using a 3-level (plus/
pre-plus/neither) scale. Image-based interpretations were integrated
with the clinical ophthalmoscopic exam to define a reference standard
diagnosis for each image. Each image was analyzed quantitatively
using a computer-based algorithm to generate a vascular tortuosity
index (TI). Diagnostic performance of each expert and the computergenerated tortuosity index were compared based on absolute
agreement and receiver operating characteristic area under the curve
(AUC), with 3-level vs. 2-level (plus/not plus) classification of plus
disease.
Results: Absolute agreement between experts and the reference
standard was 92% in the 3-level classification and improved to 96%
in the 2-level classification. Diagnostic performance of experts
measured by AUC was 0.94 in the 3-level classification, and 0.97 in
the 2-level classification. Diagnostic performance of the computerbased tortuosity index measured by AUC was 0.91 in the 3-level
classification, and 0.90 in the 2-level classification. Sensitivity
and specificity of computer-generated plus-disease diagnosis were
89% sensitivity/81%-specificity in the 3-level classification, and
100%-sensitivity/66% specificity in the 2-level classification.
Conclusions: Classifications of ROP that included a pre-plus disease
category decreased the diagnostic performance of both experts and
computer based systems. This study suggests that uncertainty in
the diagnosis of pre-plus disease hinders diagnostic performance.
Further clarification is needed in the definition of pre-plus disease to
improve the accuracy and consistency of diagnosing retinopathy of
prematurity.
Commercial Relationships: Allison R. Loh, None; Michael Ryan,
None; Katherine Abrahams, None; Esra Ataer-Cansizoglu, None;
Robison V. Chan, NIH Grant EY 19474 (F), Research to Prevent
Blindness (F), The St. Giles Foundation (F); Audina M. Berrocal,
Alcon Allergan (S); Jayashree Kalpathy-Cramer, None; Veronica
Bolon-Canedo, None; Deniz Erdogmus, None; Michael F. Chiang,
Clarity Medical Systems (S), NIH EY 19474 (F), NIH Grant EY
22387 (F), Research to Prevent Blindness (F), St. Giles Foundation
(F)
Evaluation of Diagnostic Criteria for B-Scan Ultrasonography in
Posterior Scleritis
Kristin S. Suhr1, Sarju S. Patel2. 1Ophthalmology, New York
University, New York, NY; 2Ophthalmology, Weill Cornell Medical
College, New York, NY.
Purpose: Posterior scleritis is a less common form of scleral
inflammation that can be easily misdiagnosed or overlooked. The
physician often uses a patient’s history in conjunction with exam
findings and imaging studies to make a clinical diagnosis. Exam
findings can be limited. As such, B-scan ultrasonography has become
the mainstay in diagnosing posterior scleritis. Current literature often
cites the value of 2mm and greater of scleral thickening to diagnose
the disease. However, we hypothesize that values less than 2mm on
ultrasound are often encountered in symptomatic disease. By using
the contralateral eye as a control and updating population norms of
scleral thickness, we can enhance the sensitivity of diagnosis with
B-scan ultrasound.
Methods: The retrospective case series study included patients
with active posterior scleritis and a control group. All patients were
obtained from a uveitis clinic at a tertiary-care center under one
provider. Scleral thickness was measured with ultrasound by one
trained clinician. Measurements of scleral thickness were made at
initial examination using the mean cross-section of the sclera at the
posterior pole. Additional anatomical changes were also noted, such
as “T-signs” or fluid collection between the optic nerve and sclera,
scleral nodules, and optic disc sheath distension.
Results: Eight patients (16 eyes) were included. Scleral thickness
was measured in 81.3% of patients and measurements were
reproducible. The scleral thickness in inflamed eyes was 1.94 ±
0.18mm. This was compared to normal data showing thickness of
0.94 ± 0.18mm. A significant difference was seen between the normal
mean and patients with posterior scleritis (P value < 0.00001). A
subset of patients had one unaffected eye and one affected eye which
were compared. The scleral thickness between their two eyes was
significantly different (P value of 0.03). Although scleral thickening
was evident on the images of all inflamed patients, only five eyes
presented with scleral thickness ≤ 2mm. Three eyes presented with
T-signs. No eyes showed scleral nodules. One eye showed optic disc
sheath distention.
Conclusions: Our results are consistent with our hypothesis that
B-scan ultrasound can be effectively used to observe disease at
values less than 2mm. It also opens a dialogue for updated guidelines
regarding B-scan ultrasound for posterior scleritis, ultimately to
improve the sensitivity of diagnostic testing.
Commercial Relationships: Kristin S. Suhr, None; Sarju S. Patel,
None
Assessment of Vitreous Haze using Ultra-Wide Field Retinal
Imaging
Drew Dickson, Aniruddha Agarwal, Mohammad A. Sadiq, Robin
High, Quan Dong Nguyen, Yasir J. Sepah. University of Nebraska
Medical Center, Omaha, NE.
Purpose: Conventional fundus imaging has been used to assess
vitreous haze (VH) in patients with uveitis. Ultra-wide field (UWF)
retinal imaging that uses scanning laser technology has not been
evaluated for the detection of VH. A prospective study was performed
to explore ability of UWF retinal imaging in assessing VH.
Methods: Patients with intermediate, posterior or panuveitis were
examined to assess the level of VH using slit-lamp biomicroscopy.
Colored fundus images (30°, standard 2M field) were acquired using
Carl Zeiss FF450 (Carl Zeiss Meditech, Dublin, CA) camera. Optos
UWF 200° retinal camera (Optos P200Tx, Optos, Scotland, UK) was
also used to obtain fundus images of the same eyes at the same visit
by a single photographer. Two graders, masked to the clinical data,
independently analyzed UWF fundus images for presence or absence
of VH, without quantifying the degree of VH using any scale.
These findings were then compared to the gold standard, i.e. clinical
examination, and the detection of VH using conventional Zeiss
fundus photography to determine the ability of the UWF technique
in detecting VH. Sensitivity, specificity, positive and negative
predictive values (PPV and NPV) were calculated for UWF and Zeiss
techniques in detecting VH. Agreement between the two imaging
techniques and the two observers was calculated using the weighted
κ statistic.
Results: Thirty-two eyes (16 patients) were included in the study. In
comparing the UWF imaging and the conventional Zeiss photographs
to the clinical VH assessment, the sensitivity was 0.6 and 0.8 for
Optos UWF and Zeiss photographs, respectively. Specificity was 0.92
for Optos UWF and 0.79 for Zeiss photographs. The PPV was 0.15
and 0.44, while the NPV was 0.84 and 0.95 for Optos UWF and Zeiss
photographs, respectively. The agreement between the two imaging
techniques was modest, with κ = 0.53 (range: 0.12 – 0.93). There
was a good agreement between the two observers for detection of
VH using Optos UWF images with κ = 0.64 (range: 0.31 – 0.96). The
agreement was modest for Zeiss technique with κ= 0.39 (range: 0.05
– 0.73) (Figure).
Conclusions: UWF fundus imaging using scanning laser technique
may be used to assess VH and may be employed in the management
of intermediate, posterior and panuveitis.
Figure: Graphic representation of the agreement between the graders
and imaging techniques. The values are reported by the black squares
with their confidence intervals.
Commercial Relationships: Drew Dickson, None; Aniruddha
Agarwal, None; Mohammad A. Sadiq, None; Robin High, None;
Quan Dong Nguyen, None; Yasir J. Sepah, None
Amount of residual silicone oil in vitreous cavity is significantly
correlated with axial length
Hideki Shiihara2, 1, Hiroto Terasaki1, Naoya Yoshihara1, Hiroki
Otsuka1, Toshifumi Yamashita1, Keita Yamakiri1, Shozo Sonoda1, Taiji
Sakamoto1. 1Kagoshima university, Kagoshima, Japan; 2Tarumizu
central hospital, Tarumizu, Japan.
Purpose: To determine the parameters of the eye that are
significantly correlated with the amount of residual silicone oil
remaining after most of it is removed by vitrectomy.
Methods: Nineteen eyes of 19 patients who had silicone oil
removed were studied. The day after the surgery for silicone oil
removal, B-scan ultrasonography was performed, and the residual
silicone oil droplets were observed as hyper-echoic particles in the
ultrasonographic images. The images of the vitreous cavity were
binarized, and the ratio of area of hyper-echoic particles to the total
vitreous area was quantified and named the silicone oil index (SOI).
The correlations between SOI and clinical findings were determined.
Results: The SOI was significantly and positively correlated with
the axial length (AL) and the preoperative intraocular pressure (IOP)
(AL, R=0.676, P=0.002;preoperative IOP, R=0.771, P<0.001).
Partial correlation analysis showed that the AL remained significantly
correlated with the SOI but the preoperative IOP was not (AL,
R=0.734, P=0.001; preoperative IOP, R=0.417, P=0.096). None of
the other clinical factors was significantly correlated with the SOI.
Conclusions: Considering the significant correlation between the
amount of residual silicone oil and the AL of the eye, myopic eyes
should be carefully scrutinized for residual silicone oil.
Commercial Relationships: Hideki Shiihara, None; Hiroto
Terasaki, None; Naoya Yoshihara, None; Hiroki Otsuka, None;
Toshifumi Yamashita, None; Keita Yamakiri, None; Shozo
Sonoda, None; Taiji Sakamoto, None
208 Advanced OCT applications
Monday, May 04, 2015 8:30 AM–10:15 AM
2B/3B Mile High Blrm Paper Session
Program #/Board # Range: 1307–1313
Program Number: 1307
Presentation Time: 8:30 AM–8:45 AM
Light-induced thickening of photoreceptor outer segment layer
detected by ultra-high resolution OCT imaging
Yichao Li1, Robert N. Fariss2, Ethan D. Cohen3, Haohua Qian1.
1
NEI Visual Function Core, National Institutes of Health, Rockville,
MD; 2NEI Biological Imaging core, National Institutes of Health,
Rockville, MD; 3Division of Biomedical Physics, Office of Science
and Engineering Labs, Center for Devices and Radiological Health,
Food and Drug Administration, Silver Spring, MD.
Purpose: There has been an increasing usage of OCT imaging for
functional measurements of retinal activity due to its non-invasive
property. In this study, we investigated light-induced changes in the
retinal thickness using ultra-high resolution OCT imaging methods.
Methods: C57BL/6J mice at 4-months of age were used in this study.
Dark-adapted animals were prepared under dim red light and imaged
in a dark room, whereas light-adapted mice were imaged under room
light (200 lux). Measurements were made at the same time of the day
to avoid any possible circadian rhythm-related changes. A Bioptigen
UHR system (1.6 mm/pixel axial resolution) with EDI mode was used
to capture the OCT images. Data analysis was performed in Diver 2.4
(Bioptigen, Inc.) and ImageJ.
Results: OCT images obtained from mice in room light exhibit
a prominent dark band between the RPE and photoreceptor (PR)
tip layers when compared with those obtained from the same eyes
after over-night dark-adaptation. The whole retina thickness of
light-adapted eyes is 6.8+0.7 mm longer than those measured under
dark-adapted conditions (n=13). This difference is mainly manifested
at the PR outer segments (OS), with no significant changes in either
inner retinal thicknesses or PR inner segment length. Preliminary
results from histological analysis revealed a similar change in
OS length, indicating a good correlation between the optical
measurement with OCT image and the physical distance of retinal
tissue observed histologically. Elongation of OS can be detected even
with brief light exposure, changes of 2.0+0.4 mm after 15 min of light
(n=3), and 4.1+1.0 mm after 2hr of light (n=6). Similarly, 2hr darkadaptation leads to a shortening of 4.0+0.8 mm in OS length (n=7).
After prolonged dark-adaptation, there was an elongation of OS when
compared with over-night dark-adapted eyes, with changes of 2.4+1.0
mm for 8-day darkness (n=7) and 2.2+0.6 mm for 27-day darkness,
respectively. These changes in prolonged darkness are consistent with
photostasis mechanisms of rod PR.
Conclusions: Light-induced change in the OS length is shown as an
extra dark band on OCT image. This is consistent with light-induced
fluid accumulation under sub-retinal space. If confirmed, OCT
images can be used as another non-invasive measure of retinal and
RPE function.
Adaptive-Optics SLO imaging combined with phase-variance
OCT for precise 3D localization of fluorescent cells in the mouse
retina
Robert J. Zawadzki1, 2, Pengfei Zhang1, Azhar Zam1, Eric B. Miller1,
Ravi S. Jonnal2, Dae Yu Kim3, Yifan Jian4, John S. Werner2, Marie
E. Burns1, 2, Edward Pugh1. 1Cell Biology and Human Anatomy, UC
Davis, Davis, CA; 2Ophthalmology & Vision Science, University of
California Davis, Sacramento, CA; 3Beckman Laser Institute Korea,
Dankook University, Cheonan, Korea (the Republic of); 4School of
Engineering Science, Simon Fraser University, Burnaby, BC, Canada.
Purpose: To evaluate feasibility of a mouse Adaptive Optics –
Scanning Light Ophthalmoscope (AO-SLO) and phase variance
Optical Coherence Tomography (pv-OCT) for precise 3D localization
of cells in the mouse retina in vivo.
Methods: A mouse AO-SLO system has been used to investigate
with high resolution (~1 mm) and small FOV (~150 x 150 mm) retinal
vasculature and morphology of fluorescent cells in mice. In parallel
the retinas of the same mice were imaged with custom pv-OCT
systems, allowing precise non-invasive mapping of volumetric retinal
vasculature at ~4mm resolution over a large FOV (~1.6 x 1.6mm).
The AO-SLO uses custom 0 Dpt. contact lens mounted in the
entrance pupil plane of the system to allow precise mouse positioning
and long time imaging. Several pigmented Wild Type (c57BL/6)
mice as well as mice with GFP labeled microglia (Cx3cr1GFP/+), and
neutrophils (129Sv lys-EGFP) were used as the imaging targets.
Results: The two imaging techniques offer complementary views of
mouse retinal morphology on macro (pv-OCT) and micro (AO-SLO)
scales. The figure shows mouse vasculature imaged with pv-OCT
(C,D - label free) and AO-SLO (E - Fluorescein Angiography) in
Wild Type (c57BL/6) mouse.
The retinal vasculature visualized with both pv-OCT and AO-SLO
acts as the anchor between two modalities, allowing easy registration
and precise localization of cellular structures imaged by AO-SLO to
specific retinal layers. The pv-OCT allowed label-free mapping of
retinal microvasculature that was used as a reference map for imaging
areas of interest with AO-SLO.
Conclusions: Application of two distinct retinal imaging modalities
enables precise 3D localization of AO-SLO fluorescent sources in the
retina in vivo. This capability is especially valuable for longitudinal
studies of the morphology, transformations and migration of
fluorescently tagged cells in vivo
OCT image of the same mouse retina under light- and dark-adapted
conditions.
Commercial Relationships: Yichao Li, None; Robert N. Fariss,
None; Ethan D. Cohen, None; Haohua Qian, None
Support: NEI INTRAMURAL RESEARCH PROGRAM
parafoveal region obtained by maximum intensity projection over a
depth range indicated in Fig. 1(a). We can appreciate the significant
increase in contrast especially in the microvascular structure (Fig.
1(c)).
Conclusions: Our results demonstrate, that angular compounded OA
with bidirectional DOCT allows for assessment of complementary
qualitative and quantitative flow information without compromising
vascular contrast. Both, 3D OA of retinal micro-vasculature and
absolute retinal flow information are important biomarkers in
specific ocular diseases which are not accessible with state-of-the-art
fluorescein and indocyanine green angiography.
Pv-OCT and AO-SLO fluorescein angiography (FA) from one
mouse. A. En face intensity projection image of an OCT volume. B.
B-scan from the same volume through optic disc (red arrow in A). C.
En face pv-OCT of vascular beds. D. Depth color-coded visualization
of high resolution pv-OCT (red box in C). E. Six images from AOSLO z-stack from retinal region outlined in blue in C spanning from
NFL (upper left) to just “below” the OPL (lower right).
Commercial Relationships: Robert J. Zawadzki, None; Pengfei
Zhang, None; Azhar Zam, None; Eric B. Miller, None; Ravi S.
Jonnal, None; Dae Yu Kim, None; Yifan Jian, None; John S.
Werner, None; Marie E. Burns, None; Edward Pugh, None
Support: UC Davis Research in Science & Engineering (RISE)
grant, National Eye Institute (R01 EY024239)), UC Davis National
Eye Institute Core facilities grant.
Enhanced vascular contrast with angular compounded OCT
angiography in quantitative bidirectional Doppler OCT at
400kHz
Laurin Ginner1, Daniel Fechtig1, Cedric Blatter1, Christoph Mitsch2,
Andreas Pollreisz2, Martin Gröschl2, Ursula Schmidt-Erfurth2, Rainer
A. Leitgeb1. 1Center of med. phys. and biom. eng., Medical University
of Vienna, Vienna, Austria; 2Department of Ophthalmology, General
Hospital and Medical University, Vienna, Austria.
Purpose: Bidirectional split spectrum DOCT allows performing
non-invasive optical angiography (OA) based on calculating the
speckle variance image together with quantitative absolute flow
measurement. We propose coregistration of the three-dimensional
(3D) OA image of both beams to suppress speckle and increase the
vascular contrast for in vivo retinal data.
Methods: The swept source dual beam bidirectional OCT operates at
200 kA-scans/s centered at 1050nm and uses a spinning Dove prism
that rotates both beams in the eye pupil plane for angle independent
quantitative flow assessment. A 16° OA with 800x400 lateral
sampling points is acquired in 7.8s. The implemented AC technique
registers two en-face images obtained from different viewing angles
using an angular transformation matrix. Subsequent averaging of both
speckle variance images allows increasing the vascular contrast.
Results: We measured the absolute blood flow in retinal arteries
and veins of healthy volunteers with angle-independent precision of
3% relative standard deviation in the range of 35-45ml/min. In Fig.
1(b) and 1(c) we compare enface projections of OA images of the
Fig. 1. (a) Tomogram across the fovea centralis and indicating
the depth range (red box) to calculate optical angiographies via
maximum intensity projection. (b) and (c) show enface OA images
prior and after angular compounding.
Commercial Relationships: Laurin Ginner, None; Daniel Fechtig,
None; Cedric Blatter, None; Christoph Mitsch, None; Andreas
Pollreisz, None; Martin Gröschl, None; Ursula Schmidt-Erfurth,
None; Rainer A. Leitgeb, None
Optic disc perfusion in glaucoma with optical microangiography
(OMAG)
Chieh-Li Chen1, 2, Divakar Gupta2, Joanne C. Wen2, Raghu C.
Mudumbai2, Murray A. Johnstone2, Philip P. Chen2, Karine D.
Bojikian2, Qinqin Zhang1, Yanping Huang1, Ruikang K. Wang1, 2.
1
Department of Bioengineering, University of Washington, Seattle,
WA; 2Department of Ophthalmology, University of Washington,
Seattle, WA.
Purpose: To investigate the optic disc perfusion differences in
glaucomatous and normal eyes using Cirrus 5000 HD-OCT based
OMAG.
Methods: Eyes of normal and glaucoma subjects were scanned with
a 67 kHz Cirrus 5000 HD-OCT based OMAG prototype system
(Zeiss, Dublin, CA; 245x245 raster cube scan over a 2.4x2.4mm2
area centered at the optic nerve head (ONH) region). Microvascular
images were generated from the OCT dataset by a proprietary phase
compensation method. Two layers were segmented in the ONH
region on the microvascular images using proprietary semi-automatic
segmentation software: pre-lamina layer (from the inner limiting
membrane (ILM) to the anterior surface of the lamina cribrosa
(LC)) and LC (between the anterior surface of LC and the outer
boundary of choroid). OMAG enface images for pre-lamina layer,
LC, and the entire ONH (from ILM to the outer boundary of choroid)
were generated using maximum projection. The optic disc margin
was defined using Bruch’s membrane opening and was manually
delineated on the structural enface image. Optic disc perfusion was
measured by calculating the mean flow intensity within the optic disc.
Repeated measures analyses were performed to compare optic disc
perfusion between glaucomatous and normal eyes and accounted for
cluster effect of including both eyes from the same subject. P<0.05
was considered as statistically significant.
Results: 25 eyes from 13 glaucoma subjects (mean visual field
mean deviation -4.43±5.19 dB) and 15 eyes from 8 age-matched
normal subjects were recruited. There was no significant difference
in age between normal and glaucoma subjects (normal vs glaucoma:
59.0±6.6 vs 70.9±3.7 yrs, p=0.11, t-test). Normal eyes had
statistically significantly thicker retinal nerve fiber layer thickness
compared to glaucomatous eyes (normal vs glaucoma: 85.1±13.2 vs
72.1±10.4 mm, p=0.03, t-test). Optic disc perfusion was statistically
significantly lower in glaucomatous eyes compared to normal eyes
in pre-lamina layer (normal vs glaucoma: 0.29±0.03 vs 0.26±0.03,
p=0.024) and the entire ONH (0.30±0.03 vs 0.27±0.03, p=0.022)
but not in LC (0.17±0.02 vs 0.17±0.03, p=0.79) (Figure). The
quantification value of LC may be susceptible to the limited light
penetration into deep tissue.
Conclusions: Optic disc perfusion detected by OMAG was
significantly lower in glaucomatous eyes than normal controls. Optic
disc perfusion measurement may enable detection and monitoring of
glaucoma.
Commercial Relationships: Chieh-Li Chen, Carl Zeiss Meditec
(F); Divakar Gupta, None; Joanne C. Wen, None; Raghu C.
Mudumbai, None; Murray A. Johnstone, Allergan (P), Cascade
Ophthalmics (C), Healonics (C), Ivantis (C), Sensimed (C); Philip P.
Chen, None; Karine D. Bojikian, None; Qinqin Zhang, Carl Zeiss
Meditec (F); Yanping Huang, None; Ruikang K. Wang, Carl Zeiss
Meditec (F), Carl Zeiss Meditec (P)
Support: NEI R01EY024158, Carl Zeiss Meditec Inc, Research to
Prevent Blindness
Local birefringence imaging of ocular tissue by multifunctional
Jones matrix OCT
Yoshiaki Yasuno1, 2, Satoshi Sugiyama1, 3, Young-Joo Hong1, 2, Deepa
Kasaragod1, 2, Sato Uematsu4, Masahiro Miura5, 2, Yasushi Ikuno4.
1
Computational Optics Group, University of Tsukuba, Tsukuba,
Japan; 2Computational Optics and Ophthalmology Group, Tsukuba,
Japan; 3Tomey Corporation, Nagoya, Japan; 4Department of
Ophthalmology, Osaka University Medical School, Suita, Japan;
5
Department of Ophthalmology, Tokyo Medical University Ibaraki
Medical Center, Ami, Japan.
Purpose: Polarization sensitive (PS-) OCT has been utilized to
visualize the birefringence of ocular tissue. However, PS-OCT
did not directly visualized the birefringence, but visualized phase
retardation (PR). PR is a depth-oriented cumulative effect to light
originated from the tissue birefringence. Although it was frequently
believed that the depth derivative of PR is the birefringence, it is true
only if the polarization axis orientation, i.e., the orientation of the
fibrous structure, of tissue is uniform along the depth. This condition
is satisfied with nerve fiber layer, but no for other tissues.
We present a new type of PS-OCT, so called Jones matrix (JM-)
OCT, which is capable of correctly visualizing tissue birefringence
even the axis orientation is not uniform. The limitation of PR imaging
and the advantages of birefringence are extensively discusses in
experimentally and theoretically.
Methods: A custom-built JM-OCT was utilized. This device uses
a 1060 nm probe, the measurement speed is 100,000 A-lines/s and
the depth resolution is 6.2 uμm. Volumetric scanning was performed
to cover 6 mm x 6 mm area for 6.6 s. In this method, a depthcumulative Jones matrixes are obtained first. Subsequently, depthlocalized Jones matrixes and the tissue birefringence are derived.
This raw birefringence is significantly distorted by measurement
noise. So it is corrected by using a numerical estimation algorithm
based on Bayesian theory. This process finally provides 3-D
distribution of fully quantitative tissue birefringence. 14 cases were
examined, which includes dry- and wet-AMD, polypoidal choroidal
vasculopathy (PCV), and also pathologic myopia.
Results: Fig. 1 shows an example of PCV with a huge fibrosis. Its PR
image (b) shows a non-uniform pattern in the fibrotic region, while
this region is appeared with uniformly high birefringence (green) in
the birefringence image (c). Since the fiber orientation in this region
is not uniform, the appearance of PR might be an artifact, while the
birefringence image is rational. In Fig. 2(c), the sclera of pathologic
myopic eye is appeared with two distinctive domains of birefringence
(low-blue and high-green). This domain structure is not clearly
shown in PR (b).
Conclusions: The local birefringence images clearly visualized
birefringence domains of posterior eye.
OCT (a), PR (b) and birefringence tomography (c) of PCV.
Fig. 2: A case of pathologic myopia.
Commercial Relationships: Yoshiaki Yasuno, Tomey Corp. (F),
Tomey Corp. (P), TOPCON Corp. (F); Satoshi Sugiyama, Tomey
Corp. (E); Young-Joo Hong, Tomey Corp. (F), Tomey Corp. (P),
TOPCON Corp. (F); Deepa Kasaragod, Tomey Corp. (F), Tomey
Corp. (P), TOPCON Corp. (F); Sato Uematsu, Tomey Corp. (F);
Masahiro Miura, Novartis (R); Yasushi Ikuno, Tomey Corp. (F)
Support: Japan Society for the Promotion of Science KAKENHI
2503053
Pre-laminar tissue thickness measurably affects the visualization
of the lamina cribrosa (LC)
Katie A. Lucy1, Bo Wang1, 2, Hiroshi Ishikawa1, 2, Richard A. Bilonick1,
3
, Yun Ling1, 3, Ireneusz Grulkowski4, Jonathan J. Liu4, James G.
Fujimoto4, Gadi Wollstein1, Joel S. Schuman1, 2. 1UPMC Eye Center,
Eye and Ear Institute, Ophthalmology and Visual Science Research
Center, Department of Ophthalmology, University of Pittsburgh
Medical Center, Pittsburgh, PA; 2Department of Bioengineering,
Swanson School of Engineering, University of Pittsburgh, Pittsburgh,
PA; 3Department of Biostatistics, Graduate School of Public Health,
University of Pittsburgh, Pittsburgh, PA; 4Department of Electrical
Engineering and Computer Science, Massachusetts Institute of
Technology, Cambridge, MA.
Purpose: Concrete effort has recently been made to analyze the
LC, as it is considered a primary location for glaucomatous damage.
This study evaluated the effect of pre-laminar tissue thickness on LC
visualization with OCT.
Methods: The LC region of 132 eyes (46 healthy, 29 glaucoma
suspect, and 57 glaucoma) from 105 subjects was scanned using a
prototype swept-source (SS-) OCT at a 1060nm wavelength with
a 100kHz axial scan rate. The quality of the scans were graded
subjectively using a scale of 1-5. The quality grading was based on
pore presence, clarity, and contrast. The pre-laminar tissue thickness
of each eye was quantified by measuring the distance between the
anterior pre-laminar surface and anterior laminar surface. Pre-laminar
tissue thickness was then related to the quality grading using a linear
mixed effects model, accounting for the inclusion of both eyes from
some subjects.
Results: Pre-laminar tissue thickness was 85.8±20.8mm thinner
in glaucoma suspects (p=0.0001) and 109.4±17.2mm thinner in
glaucoma subjects (p=0.0000) compared to normal (Fig. C). Prelaminar tissue thickness significantly affected the quality of LC
visualization, with thicker pre-lamina causing worse LC image
quality (p=0.0004, Fig. A, B, D).
Conclusions: Pre-laminar tissue plays an important role in the
visualization of the LC. The constraints created by pre-laminar tissue
may present difficulties in properly characterizing LC microstructure
in healthy eyes. Therefore, LC microstructure may be better used
to analyze longitudinal progression changes in individual subjects
instead of as a cross-sectional measurement to determine disease
status, until technological advances enable better LC visualization in
spite of optically scattering intervening structures.
Figure. (A) Cross-sectional scan of a healthy subject with thick prelaminar tissue (left, anterior pre-laminar surface - red line, anterior
laminar surface – teal line) shows poor LC quality (right) in C-mode.
(B) Cross-sectional scan of a glaucoma subject with thin pre-laminar
tissue (left) shows good LC quality (right) in C-mode. (C) Boxplot of
pre-laminar tissue thickness based on diagnosis and (D) scatterplot of
pre-laminar tissue thickness with ranked LC quality.
Commercial Relationships: Katie A. Lucy, None; Bo Wang, None;
Hiroshi Ishikawa, None; Richard A. Bilonick, None; Yun Ling,
None; Ireneusz Grulkowski, None; Jonathan J. Liu, Topcon (E);
James G. Fujimoto, Optovue (I), Optovue (P), Zeiss (P); Gadi
Wollstein, None; Joel S. Schuman, Zeiss (P)
Support: NIH: R01-EY013178, R01-EY011289, P30-EY008098;
Eye and Ear Foundation (Pittsburgh, PA); Unrestricted Grant from
Research to Prevent Blindness, New York, NY
Handheld Convertible Anterior/Posterior Segment Swept Source
Optical Coherence Tomography System
Derek Nankivil1, Gar Waterman1, Brenton Keller1, Anthony N. Kuo2,
Joseph A. Izatt1, 2. 1Department of Biomedical Engineering, Duke
University, Durham, NC; 2Department of Ophthalmology, Duke
University Medical Center, Durham, NC.
Purpose: Build a swept source optical coherence tomography
(SSOCT) imaging system that uses coherence revival to image
both the anterior and posterior segments with a single handheld
microelectromechanical systems (MEMS)-based optical probe.
Methods: An SSOCT system with a Mach-Zehnder topology
was constructed using an Axsun Technologies laser with a central
wavelength of 1040 nm, a 100 nm bandwidth, and a 100 kHz sweep
rate. The retinal path was matched to the reference path, giving a
conventional configuration. Coherence revival was used to encode the
anterior segment beam with a carrier frequency, effectively creating
a complex conjugate resolved extended-depth heterodyne SSOCT
system.
The diffraction-limited optical design gave a predicted lateral
resolution of 8.6 μm and a FOV of 20° for the posterior segment
system. The anterior segment system has a lateral resolution of 23.6
μm, a depth of focus of 3.6 mm, and an 11.2 mm square FOV. The
measured axial resolution for both paths was 6.9 mm.
A miniature motorized commercial translation stage and a bistable
rotary solenoid powered the motion systems that provide the ability
to adjust for patient refraction and to toggle between measurement
modes. A gimbal-less two-axis scanning MEMS micromirror was
used to control beam position. The entire probe weighs less than 1.2
lbs with a form factor of 8 x 3¾ x 3½ in.
Results: The peak sensitivity was 98 and 102 dB for the anterior and
posterior segment channels, respectively. For the posterior segment
channel, we observed a -6dB imaging range of 4.6 mm. For the
anterior segment channel, the peak sensitivity position appeared at
approximately 5 mm in depth, and the -6dB imaging range spanned
9.5 mm from 0.25 mm to 9.75 mm.
To demonstrate the applicability of dual depth SSOCT in vivo,
anterior segment and retinal images were acquired from healthy
human volunteers using bench-top mock-ups similar to the handheld
design.
Conclusions: A MEMS-based handheld dual depth SSOCT system
design that enables sequential imaging of the anterior segment and
retina was demonstrated.
246 Anterior segment imaging
Monday, May 04, 2015 11:00 AM–12:45 PM
Program #/Board # Range: 1951–1972/D0154–D0175
Contributing Section(s): Clinical/Epidemiologic Research, Cornea
Handheld SSOCT system: Isometric view (top) and side views
(bottom) illustrating the motion systems with semi-transparency and
a blue luminescent tone.
Dual Depth SSOCT Image: Anterior segment and retinal images.
B-Scan images were registered and averaged 5 times. The volume
and B-Scan acquisition times were 1.3 s and 10 ms respectively.
Commercial Relationships: Derek Nankivil, None; Gar
Waterman, None; Brenton Keller, None; Anthony N. Kuo, None;
Joseph A. Izatt, Bioptigen, Inc. (S)
Support: USAMRMC Award # W81XWH-12-1-0397, Fitzpatrick
Foundation Scholar (DN).
Program Number: 1951 Poster Board Number: D0154
Presentation Time: 11:00 AM–12:45 PM
Iridocorneal angle description in preterm patients with retcam
Tania Moron Vidal, Jesus Jimenez-Arroyo, Magdalena GarcíaHuerta, Itzel Montserrat Perez Gudino, Jesus Jimenez-Roman, Maria
A. Martinez-Castellanos. Asociación para Evitar la Ceguera en
México l.A.P “Dr. Luis Sánchez Bulnes”, México Distrito Federal,
Mexico.
Purpose: Describe morphological and anatomical characteristics of
iridocorneal angle in preterm newborns as well as correlate regular
aperture degree according to different gestational week.
Methods: Transversal, observational, descriptive, case series study.
We included 58 eyes of 29 patients.
Temporal quadrant images were taken using RetCam by an specialist
in retina and glaucoma without sedation.Image interpretation was
performed by two glaucoma specialist.
Results: From 27-28 gestational week (GW) (20.68%), we observed
in 58% closed angle; between 29 to 30 GW (34%), trabecular
meshwork was identified in 65%; whereas from 31-33 GW (31.02%),
ciliary body band was observed in 38%; finally from 34 to 36 GW
(13.39%) scleral spur was identified in 25%.
Conclusions: It is posible to correlate angular aperture with
gestational weeks.
RetCam in non-invasive method for ophtalmologic examination,
useful to study anterior chamber anatomy in vivo.
Commercial Relationships: Tania Moron Vidal, None; Jesus
Jimenez-Arroyo, None; Magdalena García-Huerta, None; Itzel
Montserrat Perez Gudino, None; Jesus Jimenez-Roman, None;
Maria A. Martinez-Castellanos, None
Iridocorneal Angle Imaging in Congenital Aniridia
Dominique Bremond-Gignac1, 2, Allan Benarous1, Henri Copin3,
Pascal Louage1, Clement Dhainaut1, Frederic Chiambaretta4.
1
Ophthalmology, South University Hospital, UPJV, Amiens, France;
2
Paris V University, CNRS Unit FR3636, Paris, France; 3Cytogenetics
and Reproduction Biology, South Hospital of Amiens, UPJV, Amiens,
France; 4Ophthalmology, University Hospital of Clermont Ferrand,
Clermont Ferrand, France.
Purpose: Congenital aniridia consists in a complex malformation of
the eye with congenital absence of iris. This genetic rare disease can
cause severe visual impairment occurring from various mechanisms
as glaucoma, limbal insufficiency and foveal hypoplasia. PAX6
gene is involved in the disease. The aim of the study is to describe
iridocorneal angle with anterior segment imaging analysis and to
evaluate glaucoma in congenital aniridia patients.
Methods: Retrospective study of 22 congenital aniridia patients,
range 2 to 61 years-old.They were evaluated with a complete
ocular examination. Patients had an anterior segment analysis
with Scheimpflug imaging (Pentacam) and Visante OCT. Also was
performed corneal topography, non contact pachymetry and macular
OCT. All images were recorded and analysed to measure the angle
and evaluate iris aplasia or hypoplasia.
Results: Of the 22 aniridia patients, OCT Visante and Scheimpflug
imaging appreciated iris aplasia or hypoplasia. Fifteen presented
an iris stub complicated of a glaucoma in 7 cases (47%) and 12
presented a treated glaucoma (54.5%). Of the twelve patients with
glaucoma, 5 had no iris stub visible to Visante OCT (41%). Of the 7
patients with no iris stub, 5 presented a glaucoma (71%). The twelve
patients were 20 years-old or more and 9 patients presented a bilateral
glaucoma (75%).Central pachymetry measurements were evaluated
for 16 patients and 14 presented a corneal thickness superior to 600
microns.
Conclusions: Congenital aniridia is a panocular disease with
developmental abnormalities, limbal stem cell deficiency and
trabeculum dysgenesia. Risk factor of glaucoma in aniridic eye
is a common complication difficult to evaluate precisely and an
evaluation of central cornea meets a an increase thickness and allows
to estimate the exact ocular pressure. Specific anterior segment
analysis is required for the follow-up of aniridia patients in order to
get a better evaluation of glaucoma risks and treatment.
Commercial Relationships: Dominique Bremond-Gignac, None;
Allan Benarous, None; Henri Copin, None; Pascal Louage, None;
Clement Dhainaut, None; Frederic Chiambaretta, None
Irido-corneal angle and fundus documentation using a novel
dual-mode imaging probe
Baskaran Mani1, 2, Shinoj VK3, Murukeshan VM3, Tin Aung1,
2 1
. Glaucoma, Singapore Eye Research Institute and Singapore
National Eye Centre, Singapore, Singapore; 2Ophthalmology, Yong
Loo Lin School of medicine, NUS, Singapore, Singapore; 3School
of Mechanical and Aerospace Engineering, Center for Optical and
Laser Engineering, Nanyang Technological University, Singapore,
Singapore.
Purpose: To evaluate a novel hand-held probe capable of dual mode
(optical and infrared) imaging of the irido-corneal angle and fundus
in a pre-clinical setting.
Methods: We have designed and developed a novel hand-held probe
integrating a miniaturized charge-coupled device (CCD) video
camera, light emitting diode (LED) light source and near infrared
light source with specifications to capture the irido-corneal angle
images and posterior segment images (Figure 1). The field of view
for the camera was set at 140° for the irido-corneal angle. This dual
mode probe was tested in a preclinical setting for imaging the angle
and fundus in rabbits and non-human primates (NHP).
Results: The irido-corneal angle (Figure 2a – in rabbits; 2b – in
NHP) and the fundus images (optic nerve head and posterior pole,
Figure 2c) were captured successfully in rabbits using contact
procedure with an ophthalmic gel interface. Approximately 100° of
the iridocorneal angle and 60° of the posterior fundus image can be
captured using the probe. The near infrared image corresponding
to the angle image could be captured by alternating the light source
without changing the probe position (Figure 3a and 3b).
Conclusions: We have successfully demonstrated the capabilities of a
dual mode novel hand-held probe for imaging the irido-corneal angle
and fundus using contact procedure. This probe can be used with the
slit-lamp light source or during surgical procedures for inexpensive
documentation of the angle and fundus images. Further refinement is
needed for improving ease of image capture and image quality.
Commercial Relationships: Baskaran Mani, Singapore Eye
Research Institute (P); Shinoj VK, Nanyang Technological
University (P); Murukeshan VM, Nanyang Technological
University (P); Tin Aung, Singapore Eye Research Institute (P)
Support: The authors acknowledge the financial support received
through NMRC (NIG09nov001), Eye Surgery and Innovative
Technologies (EyeSITe) Programme: Stratified Medicine for Primary
Angle Closure Glaucoma (NMRC/TCR/008-SERI/2013) and
A*STAR-SERC grant (No. 112 148 0003).
Effects of Ambient Light Level Variation on Spectral Domain
OCT-derived Anterior Chamber Angle Metrics in Non-Asians vs.
Asians
Ken Marion, Moritz Niemeyer, Anna Dastiridou, Brian Francis,
Srinivas R. Sadda, Vikas Chopra. DIRC, Doheny Eye Institute, Los
Angeles, CA.
Purpose: To better understand the effect of ambient light level
variation on spectral domain OCT-derived anterior chamber angle
metrics in Non-Asians vs. Asians.
Methods: Non-Asian (n=24) and Asian participants (n=24) with
open angles by gonioscopy had both eyes imaged twice at five
strictly controlled, ambient light levels at UCLA Doheny Eye Center.
Ethnicity was self-reported.
Light levels strictly controlled using a light meter at 1.0, 0.75,
0.5, 0.25, and 0 foot candle at the eye interface (fce). Each eye
was imaged 10 times for a total of 960 assessments in the study.
Zeiss Cirrus SD-OCT 5-line raster scans at IC270 irido-corneal
angle were measured twice by 2 trained, masked graders from the
Doheny Image Reading Center using customized Image J software.
Metrics: Schwalbe’s Line-angle-opening-distance (SL-AOD) and
SL-trabecular-iris-space area (SL-TISA). Delta change differences
in measurements from brightest to darkest ambient light levels were
computed. Statistical analysis completed using MedCalc v12.5.0.0.
Results: Non-Asian eye angle metrics (SL-AOD 0.443±0.114mm/
SL-TISA 0.330±0.105mm2) are significantly larger (~38%,
p=0.016) than Asian eyes (SL-AOD 0.322±0.120mm/SL-TISA
0.237±0.092mm2)(Graph 1, Table 1a). Asian eyes are more
susceptible to light variations compared to Non-Asian eyes (repeated
measures ANOVA, p=0.009). Light meter reading 1.0 to 0 fce; NonAsian Δ change SL-AOD 7.89%, SL-TISA 10.71% vs. Asian eyes
Δ change SL-AOD 13.39%, SL-TISA 16.54% (Table 1b). Excellent
Intra- and Inter-grader reproducibility/reliability (Table 1c).
Conclusions: Anterior chamber angle morphometrics are exquisitely
sensitive to changes in ambient illumination in all populations,
especially in Asian eyes. Schwalbe’s-line based ACA metrics from
the Cirrus SD-OCT could be measured with excellent intra- and
inter-grader reproducibility. These findings highlight the need to
strictly control lighting conditions in order to obtain reproducible
measurements of anterior chamber geometry.
Graph 1: Anterior Segment Angle Metrics [a] SL-AOD, [b] SL-TISA
in Non-Asian vs. Asian Populations with varied ambient light levels
Table 1: [a] Non-Asian vs Asian measurement analysis [b] Percent
change from brightest ambient room lighting to darkest [c] Grader
reproducibility/reliability
Commercial Relationships: Ken Marion, None; Moritz Niemeyer,
None; Anna Dastiridou, None; Brian Francis, None; Srinivas R.
Sadda, Carl Zeiss Meditec (C), Carl Zeiss Meditec (F), Carl Zeiss
Meditec (R), Optos (C), Optos (F), Optos (R); Vikas Chopra, None
Three-dimensional morphometrical study of the iris by swept
source anterior segment optical coherence tomography in
caucasian population
Alessandro Invernizzi1, 2, Piero Giardini1, Mario V. Cigada1,
Francesco Viola2, Giovanni Staurenghi1. 1Eye Clinic, Luigi Sacco
Hospital, University of Milan, Rho, Italy; 2Department of Clinical
Sciences and Community Health, Ophthalmological Unit, IRCCS-Cà
Granda Foundation - Ospedale Maggiore Policlinico - University of
Milan, Milan, Italy.
Purpose: to analyze in vivo three-dimensional iris morphology
(thickness/width/volume and pupil size) by swept source anterior
segment optical coherence tomography (SS-ASOCT) and to find an
eventual correlation with different influencing factors in caucasian
population.
Methods: Consecutive healthy emmetropic (refractive error
spherical equivalent between ± 3 Diopters) volunteers underwent
standardized anterior segment photography to assess iris color and
grade it. SS-ASOCT images of the enrolled eye were also obtained.
Morphometric datas of the iris comprising volume, thickness, width
and pupil size were calculated by OCT images analysis. Sectorial
variations of morphometric datas among the main 4 sectors of the
iris (superior,nasal,inferior,temporal) were assessed. Relationships
between iris morphology and iris color, subject age and gender were
studied.
Results: 135 eyes from 135 subjects (57 males and 78 females)
fulfilled the inclusion criteria and were enrolled in the study. Mean
age was 49±17 years (range 22-82). All the morphometric parameters
of the iris analyzed in the study significantly change among the
different sectors (all p<0.0001). Iris total volume and mean thickness
significantly changed depending on the color (p<0.0001 and
p=0.0384 respectively) with an increasing trend directly correlated
with iris darkening. No correlation was found between iris color and
width or pupil diameter. Age did not affect iris volume or thickness,
whereas a significant (rs=0.52) increase in iris width was found with
aging. In contrast pupil diameter significantly decreased (rs=-0.58)
with subject aging. No correlations were found between sex and
iris volume or thickness. Iris width significantly grater (p=0.007) in
males. Pupil diameter did not change among genders.
Conclusions: Morphology of the iris in caucasian population follows
constant rules in sectorial variations that can be applied to identify
pathological changes in single subjects. Iris volume and thickness
differently from iris width and pupil diameter are influenced by iris
color. Iris width and pupil size change with aging. Males have wider
iris than females.
Commercial Relationships: Alessandro Invernizzi, None; Piero
Giardini, None; Mario V. Cigada, None; Francesco Viola, None;
Giovanni Staurenghi, Heidelberg Engineering (C), Optovue (S),
Zeiss (C)
Ultrasound Biomicroscopic Analysis of Normal Human Ciliary
Body
Akari Terauchi1, Yoshifumi Okamoto1, Fumiki Okamoto1, Shinichiro
Nakano2, Tetsuro Oshika1. 1Ophthalmology, University of Tsukuba,
Tsukuba, Japan; 2Ryugasaki Saiseikai Hospital, Ryugasaki, Japan.
Purpose: To quantitatively analyze the morphology of normal human
ciliary body using ultrasound biomicroscopy.
Methods: With ultrasound biomicroscopy, we evaluated 85 eyes
of 85 normal subjects (35 males and 50 females) aged 11–86 years
(mean ± SD, 56.8 ± 20.4 years). The eyes were assessed in the 3-,
6-, 9-, 12 -o’clock quadrants relative to the center of the cornea. We
collected clinical data, including age, axial length, ciliary body length
(CBL), ciliary body thickness (CBT), anterior-chamber depth, iris
root thickness, trabecular-iris angle, and scleral-ciliary process angle.
Axial length was measured using A-scan ultrasonography.
Results: CBL and CBT tended to be greater in the superior quadrant
and decreased in the inferior quadrant, though the differences
between the four quadrants were not significant. The average CBL
and CBT showed a significant positive correlation (r = 0.40, p <
0.001). Both the average CBL and CBT were significantly correlated
with the axial length (r = 0.33, p = 0.031; r = 0.46, p < 0.01,
respectively). In addition, average CBL was significantly correlated
with anterior-chamber depth (r = 0.23, p < 0.05), trabecular-iris angle
(r = 0.29, p = 0.01), and scleral-ciliary process angle (r = 0.40, p <
0.001).
Conclusions: Ultrasound biomicroscopic imaging demonstrated that
the configuration of ciliary body is similar in all circumferences, and
highly myopic eyes have an elongated and thickened ciliary body.
The values obtained in the present study should be useful as standard
clinical references.
Commercial Relationships: Akari Terauchi, None; Yoshifumi
Okamoto, None; Fumiki Okamoto, None; Shinichiro Nakano,
None; Tetsuro Oshika, None
A Big Data Look at Lens Autofluoresence and Scattering increase
with Age in 5000+ Subjects
Zoya Hameed1, Paul Williams2, Shardendu Mishra2, Keith Ignotz2.
1
Ophthalmology, Maidstone Hospital, Kent, United Kingdom;
2
Freedom Meditech, San Diego, CA.
Purpose: The increased rate of the accumulation of advanced
glycation end products (AGEs) in the human lens of diabetics versus
healthy subjects was first described over 30 years ago and has since
been verified by several authors. AGEs exhibit fluorescence unique
to their surroundings which can be elicited following excitation.
This physiology has now been harnessed in order to recognise and
identify Non-Insulin Dependent Diabetes Mellitus (NIDDM) in the
population earlier than ever before. NIDDM onset is thought to occur
between 4-7 years prior to actual diagnosis today. The predictive
value of a high autofluorescence measurement for a diagnosis of
diabetes depends on the distribution of lens autofluorescence in a
population without known clinical disease. Our aim is to describe
the correlation between age and lens autofluorescence using readings
obtained by the Clear Path DS-120 biomicroscope and illustrate any
relationship between age and lens scatter.
Methods: Lens autofluorescence testing was carried out on the ClearPath DS 120 (v. 2.3.0.15) across 14 centres in the US over 13 months
from August 2013 to September 2014. Subjects’ age and gender were
also recorded. The scanning system is automated, takes 6 seconds to
perform from the point of fixation, and incorporates an eye tracker.
Data was de-identified, remotely pooled and analysed to establish any
relationships between age, fluorescence values and scatter.
Results: A total of 5,836 patients scans were obtained of which
47.4% were male and 52.6% were female. Subjects ranged from
8-86 years of age with a mean of 44.5 years. A strong direct linear
correlation was found between fluorescence values and age between
10-80 years. With the fluorescence value displayed as a ratio against
scatter readings we are able identify those above the 85th and 95th
percentile of the norm for a given age. With data normalised for age
and presented as a Z-score we were again able to highlight those
distinctly above normal values. Lens scattering appears to accelerate
with age possiblly due to lens damage and cataract formation.
Conclusions: Our data set confirms a strong direct relationship
between AGE autofluorescence and age when using the Clear-Path
DS 120. A expected increase in scatter with age is also demonstrated.
Further study is required to determine the affect of scatter on the
Clear-Path DS 120 scanning system as well as other physiological
characteristics.
Commercial Relationships: Zoya Hameed, Freedom Meditech
(C); Paul Williams, Freedom Meditech (E); Shardendu Mishra,
Freedom Meditech (C), Freedom Meditech (C); Keith Ignotz,
Freedom Meditech (E)
Adjusting axial length between ultrasound and partial coherence
interferometry to minimize differences in postoperative spectacle
power following cataract refractive surgery
Hang M. Tran1, 2, Jessica Taibl2, Samir I. Sayegh2. 1Western
University of Health Sciences, Pomona, CA; 2The Eye Center,
Champaign, IL.
Purpose: To demonstrate the effect of adjusting axial length
measurements of cataractous eyes on change in spectacle power and
provide a simple method to perform the adjustment.
Methods: We performed a retrospective review of axial lengths of
cataractous eyes prior to cataract surgery. A-scan ultrasound (ASU)
device, DGH 5100E, and a partial coherence interferometry (PCI)
device, IOLMaster500, measured axial lengths on the same day on
the same cataractous eye(s) of the same patient, Axial lengths based
on the PCI measurements were categorized into 3 groups: short
(<22.5mm), medium (22.5mm to < 25.0mm), and long (>=25.0mm).
Uncorrected axial length was an ASU value and corrected axial
length was an ASU value adjusted by the mean difference within
each group. Within each axial length group, the mean difference
and standard deviation (SD) between PCI and ASU were reported.
Additionally, predicted spectacle power difference between PCI and
ASU within 0.5 diopters was reported for uncorrected and corrected
axial length.
Results: The sample consisted of 126 cataractous eyes in 71 subjects
ranging from 43 to 92 years old with the average age of 68. There
were 11 short, 86 medium, and 29 long axial length eyes. The mean
difference between PCI and ASU was 0.16mm (P<0.01). The mean
difference for short axial length was 0.24mm (SD=0.071mm),
medium axial length was 0.15mm (SD= 0.22mm), and long axial
length was 0.16mm (SD=0.22mm). In the short axial length group,
uncorrected axial length showed 72.73% and corrected axial length
showed 100% spectacle power concordance within 0.5 diopters. In
the medium axial length group, uncorrected axial length showed
81.40% and corrected axial length showed 93.02% concordance.
Finally, in the long axial length group, the uncorrected axial
length showed 65.52% and corrected axial length showed 86.21%
concordance.
Conclusions: In a clinical context and with the devices used, axial
length differences between ASU and PCI appear to vary as a function
of axial length. This suggests the need for a variable adjustment
factor to complement the different A constant recommendations for
ASU and PCI made by manufacturers.
Commercial Relationships: Hang M. Tran, None; Jessica Taibl,
None; Samir I. Sayegh, None
Noninvasive Quantitative Elastography of the Cornea and the
Lens with Optical Coherence Elastography
Kirill Larin1, 2, Michael D. Twa3, Fabrice Manns4, Salavat Aglyamov5.
1
University of Houston, Houston, TX; 2Molecular Physiology and
Biophysics, Baylor College of Medicine, Houston, TX; 3School of
Optometry, University of Alabama, Birmingham, AL; 4Ophthalmic
Biophysics Center, Bascom Palmer Eye Institute, Maimi, FL;
5
Biomedical Engineering, University of Texas at Austin, Auston, TX.
Purpose: To evaluate the capability of a novel combined focused
ultrasound/air-puff and phase-sensitive optical coherence
elastography (OCE) system to assess biomechanical properties of
ocular tissues (such as cornea and the lens) in situ and in vivo in 3D.
Methods: Low-amplitude elastic deformations in mice and rabbit
ocular tissues were measured by the OCE system consisting of a
spectral-domain optical coherence tomography (OCT) combined with
focused ultrasound (lens excitation) and air-puff (cornea excitation)
systems used to produce a transient force on the tissue surface. The
amplitude, temporal profile, and the speed of the deformations were
used to reconstruct tissue biomechanical properties using novel
analytical models. Gold standard uniaxial compressional tests were
used to validate the OCE data.
Results: The OCE measurements in rabbit lens showed that the
amplitude and the relaxation rate of the displacements (and, thus,
Young’s modulus and shear viscosity) of the young lenses were
significantly larger than those of the mature lenses, indicating a
gradual increase of the lens stiffness with age (2.5 kPa and 7.4 kPa,
respectively), Fig. 1. 3D visualization of the elastic wave propagating
in rabbit corneas shows the obvious velocity difference in normal
(0.79±0.050 m/s) and CLX (2.00±0.23 m/s) corneas (Figure 2). Then,
the frequency analysis allowed depth-resolved analysis of cornea
biomechanical properties clearly demonstrating the biomechanical
differences of the corneal layers. Finally, the stress-strain
measurements using uniaxial mechanical tests confirmed the results
obtained by the OCE system.
Conclusions: The results demonstrate that the OCE system can be
used for noninvasive analysis and quantification of cornea and lens
biomechanical properties in 2D (lens) and in 3D (cornea) and as a
function of age or therapy (e.g. CLX procedures).
Figure 1. (a) Young’s modulus and (b) shear viscosity modulus
of young (n=3) and mature (n=4) lenses estimated from US-OCE
measurements.
Figure 2. 3D visualization of the elastic wave propagating in normal
(a) and CLX (b) rabbit corneas
Commercial Relationships: Kirill Larin, None; Michael D. Twa,
None; Fabrice Manns, None; Salavat Aglyamov, None
Support: NIH grants R01EY022362 and R01EY014225
Automatic segmentation of corneal endothelial cells by a genetic
algorithm
Fabio Scarpa, Alfredo Ruggeri. Department of Information
Engineering, University of Padova, Padova, Italy.
Purpose: To develop a fully automatic, reliable cell detection
system for corneal endothelium images from clinical specular
microscopy. The system should eventually allow the measurement
of the morphometric parameters that provide an objective clinical
assessment of corneal endothelium (density, polymegethism,
polymorphism). To this purpose, a new cell segmentation method
based on a genetic algorithm was developed.
Methods: Thirty images of corneal endothelium were acquired with a
specular endothelial microscope (SP-3000P, Topcon Co., Japan) from
both healthy and pathological subjects. The region of interest covered
an area of about 0.1 mm2, including on average 220 cells. Each cell
contour is detected using a genetic algorithm, a method for solving
optimization problems based on a selection process that mimics
biological evolution. It randomly modifies individuals from the
current population to produce the children for the next generation and
over successive generations the population evolves toward an optimal
solution. For the specific application, a small set of vertexes (i.e.
individuals) forming regular hexagons is used as starting population.
At each step, the location of each vertex is randomly modified. Thus,
the initial regular hexagons evolve into polygons with possibly
different number and positions of vertexes, with each vertex
reliability evaluated by considering both its correspondence with the
actual image (pixels intensity) and the regularity of the polygons.
The entire population of vertexes forms a set of polygons that fit
the underlying cells contours. The number of polygons is step-wise
increased so as to iteratively detect the contour of all the cells in the
image (Figure 1). From these contours the morphometric parameters
of clinical interest can then be easily computed.
Results: The preliminary results obtained for the contour detection
(see example in Figure 1) show that the segmentation of the corneal
endothelial cells provided by the proposed method is in good
agreement with the ground truth segmentation, obtained with a
careful, time-consuming manual analysis.
Conclusions: The proposed totally automatic algorithm appears
capable of reliably obtaining the cell contour in regions containing
also hundreds of cells. This will allow the straightforward
computation of all the important morphometric parameters used in
clinical practice.
Figure 1: Representative example of the cell segmentation’s evolution
Commercial Relationships: Fabio Scarpa, None; Alfredo Ruggeri,
None
Comparison of different endothelial cell analyzing software
systems in evaluation of endothelial cells of cultured human
donor corneas
Ingo Schmack1, 2, Irina Voehringer1, Brigitte Erber1, Seda
Ballikaya1, Georg Toszkowski3, Norbert Dahmen3, Gerd Auffarth1.
1
Ophthalmology, University of Heidelberg, Heidelberg, Germany;
2
Ophthalmology, University of Frankfurt, Frankfurt/Main, Germany;
3
University of Applied Sciences, Krefeld, Germany.
Purpose: To evaluate the accuracy and efficacy of a new endothelial
cell analyzing system (REA; robin GmbH, Germany) in comparison
to a well-established endothelial cell image analyzing software
(NAVIS, Nidek Technologies, Japan).
Methods: Organ-cultured human donor corneas were evaluated by
light microscopy after 6 hours of deswelling (CorneaJet, Eurobio,
France). Endothelial cells counts were assessed with a computerbased fixed-frame method using automated and semi-automated
algorithms of two endothelial cell analyzing systems (REA; robin
GmbH, Germany versus NAVIS, Nidek Technologies, Japan). With
the automated program, the computer software recognized single
cells and estimated endothelial cell density based on the number
of identified cells and their areas. For semiautomatic evaluation,
the observer marked individual endothelial cells and the software
subsequently calculated endothelial cell densities.
Results: Based on endothelial cell image quality, both systems
tended to show significant differences between the automated and
semi-automated detected endothelial cell counts. In the majority,
automated detected endothelial cell densities were lower with
NAVIS software and higher using REA software compared to semiautomated endothelial cell analysis (range: 10 to 855 cells). Semiautomated cell counts were comparable between both systems.
Conclusions: The new REA system provides an easy, efficient and
reliable approach in image acquisition and endothelial cell analysis.
However, semi-automated algorithms tend to be more accurate
compared to automated algorithms in both technoligies. Therefore,
manual correction of automated acquired endothelial cell images is
still strongly recommended in evaluation of human donor corneas.
Commercial Relationships: Ingo Schmack, None; Irina
Voehringer, None; Brigitte Erber, None; Seda Ballikaya, None;
Georg Toszkowski, None; Norbert Dahmen, None; Gerd Auffarth,
None
Methods: All SM done in 2013 at the Eye Clinic Dr. Fernando
Abib, Curitiba, Brazil. The type of the CSM was determined by the
algorithm (Figure). NC CSM (CSO, Italy) and C CSM (BioOptics,
USA) were utilized. SM results were calculated by considering the
CSM endothelial sampling with a minimum number of endothelial
cells counted in each sampling process, for orientation of Reliability
Indexes of the Cells Analyzer USA Patent software program (Technicall,
Brazil). The results are reported in terms of gender, number of
examinations performed with NC and C CSM, indications and
diagnoses, utilizing descriptive statistics.
Results: 879 patients, 634 (72%) female and 245 (38%) male. 606
(69%) SM with NC CSM and 273 (31%) SM with C CSM. 19 out
of 606 (3%) designed with NC CSM had to be performed with C
CSM, the NC CSM, as it was not able to capture endothelial images.
Indications for SM with NC and C CSM are listed in the Table 1:
The most frequent with NC CSM: refractive evaluation, contact lens
wearers and pre cataract surgery; The most frequent with C CSM:
Fuchs Endothelial Dystrophy, penetrant keratoplasty and cornea
guttata.
Conclusions: With the proposed algorithm, the NC CSM is not
enough to perform the SM in all the patients, as the C CSM was
required for SM in 31% of the patients with clinical diseases and
complex situations. It is recommended that cases of endothelial
mosaic Irregular or Chameleon Pattern be assigned for the
performance of SM with C CSM, as this type is a more objective SM
for the results and versatile for carrying out the examinations.
Proposed methodology for the examination of the Corneal
Specular Microscopy and the profile of the performed
examinations
Fernando C. Abib. 1Anatomy, Federal University of Parana, Curitiba,
Brazil; 2Cornea, Clinica de Olhos Dr. Fernando Abib, Curitiba,
Brazil.
Purpose: Non-Contact (NC) Corneal Specular Microscope (CSM)
acquire endothelial mosaic (EM) images when the patient sets the
target light, each model of this type of CSM has fixed positions.
Contact (C) CSM (CSM) can acquire EM images of any area of
the endothelial surface by sliding the objective lens on the cornea.
The EM according to the author can be classified into three models:
Regular Pattern, Irregular Pattern or Chameleon Pattern. Considering
the different patterns and the way the NC and C CSM perform the
Specular Microscpy (SM), this study is in order: 1 - Present algorithm
to guide the selection of the best type of CSM to perform the SM ; 2
- Know the amount of SM done with NC and C CSM, and the related
information for each of these types CSM.
Commercial Relationships: Fernando C. Abib, Fernando C Abib
(P)
Clinical grading of corneal subbasal nerve tortuosity
Alfredo Ruggeri1, Enea Poletti1, Neil S. Lagali2. 1Dept of Information
Engineering, University of Padua, Padua, Italy; 2Eye Clinic,
Linkoping, Sweden.
Purpose: Corneal subbasal nerve tortuosity (CNT) estimated from
confocal microscopy is becoming a widely used clinical measure.
Since a standardized definition is still lacking, we investigated the
clinical perception of CNT by measuring expert graders’ agreement
when assessing several different images. Their average response
may also represent a reliable ground-truth reference for clinically
assessing CNT.
Methods: A set of 30 images with a wide range of perceived NT was
assembled from images acquired at 7 research centers (Auckland,
Boston, Linköping, Manchester, Oslo, Rostock, Sydney). 7
experienced graders were asked to sort the 30 images by increasing
NT, using a computer-aided image comparison tool that uses the
human grader assessment to order the set from lowest to highest NT.
Graders agreement was assessed by the Spearman Rank Correlation
Coefficient (SRCC) among their orderings.
Experiment 1. No specific instruction about CNT was provided. Most
of the graders reported that they found it difficult to comparatively
assess many small-range directional changes vs. few long-range ones.
Experiment 2. 4 specific definitions of CNT were proposed: average
(T1) or maximum (T2) of short-range direction changes; average
(T3) or maximum (T4) of long-range direction direction changes (see
examples in Fig. 1). The 7 graders were asked to order again the 30
image using in turn each of the 4 definitions.
Results: All results from experiment 2 are better than the one from
experiment 1 (Tab. 1). For the same type of CNT definition (shortrange or long-range), maximum (T2 and T4) scored better than
average (T1 and T3).
Average SRCCs between T1 and T2 and between T3 and T4
orderings were all >0.94, whilst between T1 and T3 and T2 and T4
were all <0.40, suggesting high correlation between average and
maximum and high independence between short and long-range
directional changes.
Conclusions: The clinical perception of CNT appears to have two
distinct forms, characterized by either short or long-range directional
changes. Both types of tortuosity need to be considered when
assessing CNT. We now need to investigate to what degree different
cornea pathologies exhibit one or the other form in their subbasal
nerve patterns.
Results: 11 patients, (7 males and 4 females, mean age 68 years;
range 22 – 74 years) were included. Only vessels containing red
blood cells were visible using colour or red free images. Additional
vessels not containing cells were detected using confocal and
angiography (mean diameter 10-20mm); often adjacent to larger
venules.
Conclusions: Not all vessels in the cornea carry or only
intermittently carry red blood cells. Imaging of CoNV complexes
using confocal and angiography allowed the differentiation between
vessels with red blood and other cell trafficking, and vessels with
acellular flow. It is unclear whether these latter ‘plasma’ vessels are
dual functional and serve as lymphatic type vessels. These vessels
may be of importance in mediating inflammation and rejection.
Fig. 1. Images with: low T1-T2 and high T3-T4 (a); high T1-T2 and
low T3-T4 (b).
Table 1. Average SRCC among graders’ ordering for the various
experiments.
Commercial Relationships: Alfredo Ruggeri, None; Enea Poletti,
None; Neil S. Lagali, None
Lymphatics and other vessels in human cornea
Vito Romano1, Bernhard Steger1, Yalin Zheng2, Colin E. Willoughby2,
Sajjad Ahmad2, Stephen B. Kaye2. 1Royal Liverpool University
Hospital, Liverpool, United Kingdom; 2UNiversity of Liverpool,
Liverpool, United Kingdom.
Purpose: To investigate corneal neovascular complexes (CoNV) in
vivo for the presence of lymphatic vessels.
Methods: Patients with clinically recognizable corneal
neovascularisation (CoNV) undergoing corneal angiography for
fine needle diathermy were included. Each CoNV complex was
imaged using slit lamp biomicroscopic photography, confocal, and
fluorescein and indocyanine green angiography (FA and ICGA).
Fluorescein dye leakage and uptake from the extravascular to
the intravascular space was investigated determined using repeat
angiography and subtraction analysis. Vessel parameters (area,
diameter, branching, tortousity, flow) were analysed using in house
software.
Commercial Relationships: Vito Romano, None; Bernhard
Steger, None; Yalin Zheng, None; Colin E. Willoughby, None;
Sajjad Ahmad, None; Stephen B. Kaye, None
Determining the position of limbus corneae from anterior eye
surface topography
Alejandra Consejo, D Robert Iskander. Wroclaw University of
Technology, Wroclaw, Poland.
Purpose: To demarcate the limbal region from a profilometry
measurement of an anterior eye surface.
Methods: Traditionally, the limbus has been assumed to coincide
with the HVID or the, white-to-white (W2W) radius estimated from
2D “en face” intensity images. Recently, a new technology based on
the principle of profilometry, which has the potential of measuring
the corneo-scleral topography extending the acquired area up to 20
mm diameter far beyond the limbus, has become available. Studying
the 3D corneo-scleral topography is possible using this technology.
Of interest was whether the 3D anterior eye scleral topography could
be utilized to demarcate the limbal area based on local curvature.
Raw 3D profilometry data was fit with the 8th radial order Zernike
polynomial expansion, from which semi-meridional curvature was
calculated. This local curvature was analytically calculated using the
first and the second derivative of Zernike polynomials. Changes in
local curvature sign were used to demarcate the limbus.
Results: In some of the studied cases no change was observed in the
sign of local curvature in the temporal side of the anterior corneoscleral topography leading to incomplete limbus demarcation. Also,
the topography-based estimation of the limbus location did not
always correspond to the limbus estimated from the changes apparent
in the intensity images (i.e., HVID, W2W), see Fig 1 and 2. It is
noticed that the limbus demarcation is specific to individual’s anterior
eye, indicating large varieties of scleral topography.
Conclusions: A method of limbus demarcation based on the Zernike
polynomial expansion of the total anterior eye topography has been
proposed. It was noticed that topography-based limbus demarcation
is not a trivial task. There are no standards. Moreover, the results
depend on the limbus definition (in our case change in sign of local
curvature) as well as the technique used for its estimation. With more
advanced technologies becoming readily available, one is able to
choose a specific definition of limbus suitable for his or her particular
purposes. The proposed topography-based estimation of the limbus
location can be particularly attractive for scleral contact lens design
and fit.
Fig1.Example of the estimated limbus position indicated by
overlapping circles. The change in the inferior region is related to
lower tear meniscus.
Fig2.Corresponding to Fig 1 curvature of the anterior surface of the
eye
Commercial Relationships: Alejandra Consejo, None; D Robert
Iskander, Eaglet EYE (F)
Support: Marie Curie ITN Grant, AGEYE, 608049
Image-Guided Modified Deep Anterior Lamellar Keratoplasty
(DALK) Corneal Transplant Using Intraoperative Optical
Coherence Tomography
Yuankai Tao, Michael LaBarbera, Justis P. Ehlers, Sunil K.
Srivastava, William J. Dupps. Cole Eye Institute, Cleveland Clinic,
Cleveland, OH.
Purpose: Deep anterior lamellar keratoplasty (DALK) has become
a popular alternative to penetrating keratoplasty (PK). One
conventional DALK technique (modified ‘big bubble technique’)
begins by performing a trephination of approximately 80% stromal
thickness, as measured by pachymetry. These procedures are
inherently difficult and intraoperative rates of Descemet’s membrane
perforation between 4-39% have been reported. Optical coherence
tomography (OCT) provides high-resolution images of tissue
microstructures in the cornea, including Descemet’s membrane, and
allows quantitation of corneal layer thicknesses. Here, we use crosssectional OCT measurements of corneal thickness during surgery
and a novel micrometer-adjustable biopsy punch to precision-cut the
stroma down to Descemet’s membrane.
Methods: Modified DALK procedures were simulated on cadaveric
porcine eyes using a prototype microscope-mounted intraoperative
OCT (iOCT) system for guidance of corneal stroma dissection.
The micrometer-adjustable biopsy punch consists of a blunt-tipped
cannula attached to a manual micrometer with 1 μm precision and
a disposable 1 mm diameter biopsy punch. Corneal thicknesses up
to the pre-Descemet’s space measured on iOCT were scaled using a
mean measured corneal refractive index and used to set the biopsy
punch cut-depth. Once a dissection plane was established, a blunt
cannula was used to inject air or viscoelastic to further dissect across
the plane, creating a space between Descemet’s membrane and the
corneal stroma (Fig. 1).
Results: iOCT-guided modified DALK procedures were performed
on 47 cadaveric porcine eyes by non-experts and achieved a
perforation rate of ~5% with a mean corneal dissection time <18
minutes. The procedure was also successful performed on a human
donor eye without perforation.
Conclusions: iOCT-guided modified DALK allows blind
dissection of corneal stroma down to Descemet’s membrane. iOCT
measurements of corneal thickness and use of a micrometer-adjusted
biopsy punch provides reproducible cut-depths with reduced surgical
complexity and corneal endothelium perforation rates. Our data
shows the potential for iOCT-guided precision anterior segment
surgery without variability as a result of tremor and improvements to
standard clinical care.
applied. A zoom-in on aligned interference peaks reveals that axial
length differs between different scans due to uncompensated eye
motion. After the signal was re-interpolated using PMI tracked eye
position, scans show good alignment. (cf. Figure 1C).
Conclusions: By using a combined PMI/OLCR modality, eye motion
artefacts in the OLCR signal could be compensated, improving
thereby the reproducibility of axial length measurement. The
proposed method for motion compensation could be applied to other
devices such as optical coherence tomography for which eye motion
compensation is critical.
Commercial Relationships: Yuankai Tao, Cleveland Clinic (P);
Michael LaBarbera, None; Justis P. Ehlers, Cleveland Clinic
(P); Sunil K. Srivastava, Cleveland Clinic (P); William J. Dupps,
Cleveland Clinic (P)
Eye motion compensation for accurate ocular biometry with
a combined Purkinje mirror imaging (PMI) and optical low
coherence reflectometry (OLCR) system
Boss Daniel1, Pavel Zakharov1, Andrew Nolan2, Michael C. Mrochen1,
2 1
. IROC Science to Innovation AG, Zürich, Switzerland; 2Clearsight
Innovation Ltd, Dublin, Ireland.
Purpose: Successful cataract surgery relies on accurate preoperative
ocular biometric measurements based on which an appropriate
intraocular lens (IOL) can be calculated. One important ocular
parameter required to design a correct IOL is the axial length of the
eye.
Over the last years, optical low coherence reflectometry (OLCR) has
become the standard of care for axial length measurements in cataract
surgery due to its high precision and accuracy.
However, because of the finite scanning time in OLCR, eye motions
occurring during the measurement can induce motion artefacts.
A simultaneous measurement of the eye position would allow to
compensate for eye motion artefacts in OLCR and hence improve the
accuracy of measured axial length.
Methods: A combined Purkinje mirror imaging (PMI) and OLCR
setup was developed. The PMI modality allows for accurate anterior
segment biometry. Additionally, the eye tracking capability of PMI
allows to register eye motions during an OLCR measurement.
To verify the eye motion compensations with PMI, a model eye was
mounted onto translational stages and typical eye motion sequences
were applied while OLCR measurements were performed.
Results: Figure 1A shows a reconstructed OLCR envelope signal
from a model eye with interference peaks at the anterior and posterior
corneal surface (P1,P2), anterior and posterior lens surface (P3,P4)
and the retina (RET). Figure 1B shows a 2-D plot of 26 consecutive
OLCR scans that were obtained while an eye motion sequence was
Figure 1 Motion compensation in OLCR scans. A) Interference
envelope signal from an OLCR scan in a model eye B) 2-D plot
of 26 consecutive OLCR scans from a model eye when typical
eye movements were applied. C) Aligned scans after motion
compensation.
Commercial Relationships: Boss Daniel, Clearsight Innovations
LTD (C), IROC Science to Innovation AG (E); Pavel Zakharov,
Clearsight Innovation LTD (C), IROC Science to Innovation AG
(E); Andrew Nolan, Clearsight Innovation LTD (E); Michael
C. Mrochen, Clearsight Innovation LTD (C), IROC Science to
Innovation AG (E)
Model eye measurement with novel device combining Purkinje
reflections and OLCR
Andrew Nolan1, Boss Daniel2, Pavel Zakharov2, Michael C. Mrochen1,
2 1
. ClearSight Innovations, Dublin, Ireland; 2IROC Science, Zurich,
Switzerland.
Purpose: Accurate biometry measurements of the pre-op eye are
important for the refractive outcomes of cataract surgery. A novel
ocular biometer consisting of a Purkinje imaging method and an
optical coherence device has been developed for this purpose. We
hypothesise that the new ocular biometer can produce highly accurate
and reproducible measurements of physical model eyes parameters.
Methods: The combination of Purkinje imaging and Optical Low
Coherence Reflectometry allows measurement of the curvatures of
anterior and posterior cornea, anterior and posterior lens, and internal
ocular distances from anterior cornea to retina. The device was used
to measure 5 model eyes with parameters representative of typical
pre-operative cataract eyes.
The mean radius of anterior and posterior cornea, anterior and
posterior lens, central corneal thickness (CCT), anterior chamber
depth (ACD), lens thickness (LT) and axial length (AL) were
measured for each model eye. These parameters had been verified
by means of independent measurements. The average of the
absolute value of the difference, Δ, between these two measurement
methods across all measurements of all model eyes is presented.
The variability for each parameter was assessed across multiple
measurements of each model eye.
Results: All parameters could be reconstructed and measured by
the new ocular biometer. The differences between the parameter are
below clinical relevant parameter. Table 1 show the variability across
multiple measurements of each of the 5 model eyes, with the average
deviation from independent measurement values.
Conclusions: Our results demonstrate that the new ocular biometer
is capable of measuring model eyes with parameters representative
of pre-op human eyes with high accuracy and repeatability. Clinical
investigations are ongoing to evaluate the reproducibility of the
device in human eyes.
Table 1. Model eye measurement results
Commercial Relationships: Andrew Nolan, ClearSight Innovations
(E), ClearSight Innovations (I); Boss Daniel, Clearsight Innovations
(C); Pavel Zakharov, ClearSight Innovations (C); Michael C.
Mrochen, ClearSight Innovations (E), ClearSight Innovations (I)
Scanning Laser Ophthalmoscope Imaging is a diagnostic
predictor of Dry Eye
Mark Espina1, Cheryl A. Arcinue1, Natalia Camacho1, Mostafa Alam1,
Qisheng You1, Maria Laura Gomez1, Dirk-Uwe G. Bartsch1, Lingyun
Cheng1, Natalie A. Afshari2, William R. Freeman1. 1Ophthalmology,
UCSD-Jacobs Retina Center, La Jolla, CA; 2Ophthalmology, UCSD,
San Diego, CA.
Purpose: To evaluate a unique reticular pattern noted in scanning
laser ophthalmoscopy (SLO) best seen in the multi-color SLO
imaging and was felt to be an objective indicator of corneal dryness.
Methods: Fifty-one (51) eyes were scanned using the Heidelberg
Spectralis SLO. Multicolor, Blue reflectance and Green reflectance
images were taken using three different wavelengths (486 nm, 518
nm, 815 nm). Tear break up time (TBUT), corneal staining and
meibomian gland grading were then evaluated by masked examiner
on the same day. Presence of corneal reticulations noted in any of the
imaging modalities as evaluated by three observers separately and
dry eye as dictated by the three parameters were then analyzed using
the Pearson Correlation Coefficient test.
Results: Fifty-one (51) eyes of 26 patients were included in this
study. Fifty (50%) percent were females; the mean age was 78
years old (57- 91). The reticular pattern predicted break up time and
corneal staining. The strongest predictor was multicolor (p value
<0.0001 with TBUT and p value 0.0007 with corneal staining). Green
reflectance also had significant correlation (p value <0.0001 with
TBUT and p value 0.0047 with corneal staining). Blue reflectance
had the least correlation (p value 0.0029 with TBUT and 0.0715 with
corneal staining). There was no association with meibomian gland
dysfunction. This suggests the reticular SLO pattern correlates with
ocular surface abnormalities and rapid evaporation in dry eye.
Conclusions: This is a newly described SLO phenomenon. This is
a reticular reflex that appears to be originating from the tear film
covering the corneal surface. This reticulation appears to be a very
rapid and objective definable measure to characterize corneal dryness.
Commercial Relationships: Mark Espina, None; Cheryl A.
Arcinue, None; Natalia Camacho, None; Mostafa Alam, None;
Qisheng You, None; Maria Laura Gomez, None; Dirk-Uwe G.
Bartsch, None; Lingyun Cheng, None; Natalie A. Afshari, None;
William R. Freeman, None
Fully-automatic segmentation of conjunctival blood vessels
Pedro Guimaraes1, Luisa Frizziero2, Jeffrey C. Wigdahl1, Edoardo
Midena2, Alfredo Ruggeri1. 1Department of Information Engineering,
University of Padova, Padova, Italy; 2Department of Ophthalmology,
University of Padova, Padova, Italy.
Purpose: The conjunctiva is one of the few places in the human body
where one can easily and directly observe the microvasculature. In
this work, we propose a novel fully-automatic approach to segment
the conjunctival blood vessels.
Methods: Conjunctival blood vessels can be imaged noninvasively
using slit-lamp microscopy. In this work, the conjunctiva of 10
patients was imaged using a digital camera attached to a SL 990
slit lamp microscope (C.S.O., Scandicci, Italy). For each patient, 4
images of 960x1280 pixels with a magnification of 10x were acquired
– one image per region (inferior, superior, nasal, and temporal).
When imaging the conjunctiva, other structures such as the iris or
eyelashes may also be present in the image. For this reason, a region
of interest (ROI) is automatically defined, where the segmentation is
performed. As a preprocessing step, each image undergoes top-hat
filtering. This method has been shown to correct for different imaging
artifacts such as uneven illumination and contrast.
Vessel segmentation is achieved using the phase-congruency
parameter, a dimensionless quantity computed for each image pixel
and invariant to contrast and scale. It measures the agreement of
the phase of the Fourier components of an image, with the same
relevance being given to all frequency components, independent of
gradient magnitude. The phase congruency is thresholded to obtain a
binary image. Finally, fragmented vessel segments are connected by
morphological dilation of the resulting binary images, while isolated
small segments are erased.
Results: All images could be segmented with the proposed algorithm.
Figure 1 shows two examples of the results, where it appears that
most blood vessels are correctly traced. The time required to analyze
a single image, using a prototype developed in MATLAB language,
was 1.83 ± 0.04 s (mean ± standard deviation).
Conclusions: The proposed algorithm seems capable of correctly
tracing the conjunctival blood vessels. This approach opens the way
to fully-automatic analysis of microvasculature, which may provide
important clinical insight for several pathologies.
Figure 1: Conjunctiva images with automatically segmented blood
vessels marked in red.
Commercial Relationships: Pedro Guimaraes, None; Luisa
Frizziero, None; Jeffrey C. Wigdahl, None; Edoardo Midena,
None; Alfredo Ruggeri, None
Support: Marie Curie grant from the European Commission in the
framework of the REVAMMAD ITN (Initial Training Research
network), Project number 316990.
Automated Assessment of Hemodynamics in the Conjunctival
Microvascular Network
Maziyar M Khansari, Justin Wanek, Nicole Camardo, Mahnaz
Shahidi. Ophthalmology & Visual Sciences, University of Illinois at
Chicago, Chicago, IL.
Purpose: Conjunctival microcirculation is accessible for direct
visualization and can provide information about microvascular
hemodynamics properties of the eye and other organs of the
body. This study reports a method for automated assessment of
hemodynamics in the conjunctival microvascular network.
Methods: Multiple image sequences of the conjunctival
microcirculation were acquired in one eye of 10 healthy human
subjects (age; 48 + 8 years). In 5 subjects, the same area of the
conjunctiva microcirculation was imaged repeatedly. Variance
filtering was used to automatically segment vessels in which red
blood cell motion was detectable. Diameter (D) and axial blood
velocity (V) were measured and blood flow (Q) and wall shear
rate (WSR) were calculated from D and V. These measurements
were categorized into 4 diameter (<15 μm, 15-24 μm, 25-34 μm,
>34 μm) and 4 velocity groups (<0.2 mm/s, 0.2-0.4 mm/s, 0.4-0.6
mm/s, >0.6mm/s). Measurement repeatability was assessed by the
standard deviation of repeated measurements. General linear model
multivariate analysis was performed to determine the effects of D
(or V) group and vessel type on D, V, Q and WSR. Significance was
accepted at P < 0.01 to correct for multiple comparisons.
Results: Compiled data yielded D and V measurements in 75
arterioles and 240 venules. The largest number of measurements were
obtained in the 15-24 μm and 25-34 μm D groups (for venules and
arterioles), and in the < 0.2 mm/s and 0.2-0.4 mm/s (for arterioles)
and > 0.6 mm/s (for venules) V groups. In arterioles and venules, D
and Q were significantly different among D groups (P ≤ 0.005), but
V and WSR were similar among D groups (P ≥ 0.03). As expected,
in both arterioles and venules, V, Q and WSR were significantly
different among V groups (P < 0.001), but D was similar among V
groups (P > 0.06). The repeatability of D and V measurements were
1 μm and 0.09 mm/s, respectively. In both arterioles and venules, V
was not linearly correlated with D (P = 0.2), while Q was linearly
correlated with D (P < 0.001). WSR was linearly correlated with D in
arterioles (P = 0.007), but not in venules (P = 0.03).
Conclusions: An automated method for quantitative and
comprehensive assessment of hemodynamics of the conjunctival
microvascular network was demonstrated. This method may be
useful for detection of microvascular hemodynamic alterations due to
systemic and ocular diseases.
Commercial Relationships: Maziyar M Khansari, None;
Justin Wanek, None; Nicole Camardo, None; Mahnaz Shahidi,
US8801183 (P)
Support: NIH EY001792, Research to Prevent Blindness
Relationship between bulbar conjunctival hemorheology and
metrics of cardiac function in patients with heart failure
Bruce I. Gaynes1, Justin Wanek3, Nicole Camardo3, Groo Vicki2,
Mayank Kansal4, Chris Gans4, Mahnaz Shahidi3. 1Ophthalmology,
Loyola University Chicago, Maywood, IL; 2Pharmacy Practice,
University of Illinios, Chicago, IL; 3Visual Sciences, University of
Illinois, Chicago, IL; 4Cardiology, University of Illinios, Chicago, IL.
Purpose:
Imaging of the human bulbar conjunctival microvasculature provides
a novel non-invasive means to assess in-vivo hemorheology in
relation to systemic disease influences.The purpose of this study was
to assess the relationship between bulbar conjunctival hemorheology
and metrics of cardiac function in subjects with congestive heart
failure (CHF).
Methods: The study employed a cross-sectional assessment of
bulbar conjunctival hemorheology including diameter, venular axial
blood velocity (Vax) and flow (Q) measured by a prototype imaging
system on conescutive patients undergoing cardiac catherization for
right or left sided heart failure. Following ocular imaging, cardiac
catheterization and echocardiography were performed to evaluate
central venous pressure, pulmonary capillary wedge pressure,
cardiac output, cardiac index, ejection fraction and pulmonary
artery and right ventricle pressure. A Pearson correlation coefficient
was computed to assess the relationship between conjunctival
hemorheology and metrics of cardiac function.
Results: Five subjects were enrolled in the study, age range 2461 years. Data was pooled for analysis. Conjunctival Vax directly
correlated with vessel diameter (R = 0.5, p<0.001). Conjunctival
Q was inversely correlated with cardiac ejection fraction (R = -0.9,
p = 0.02). Central venous pressure and right systolic ventricular
pressure appeared to demonstrate a non-linear relationship with
conjunctival Q.Two subjects demonstrated significant segmentation
of conjunctival Q characterized by anomalous rouleaux formation.
Conclusions: Microvascular axial blood velocity is directly
proportional to vessel diameter suggesting a proportional reduction
in flow resistance with increased vessel diameter. It is unclear if the
relationship between Q and ejection fraction is a consequence of local
microvascular abnormalities due to CHF or hemodynamic influences
of cardiac function. The relationship between cardiac ejection
fraction and Q, may however prove useful in prognostic serial
assessment of CHF. Examination of microvascular nouleaux may
be of benefit in optimizing pharmacotherapy related to erythrocyte
aggregation.
Commercial Relationships: Bruce I. Gaynes, Patent # US8801183
(P); Justin Wanek, Patent #US8801183 (P); Nicole Camardo, None;
Groo Vicki, None; Mayank Kansal, None; Chris Gans, None;
Mahnaz Shahidi, Patent #US8801183 (P)
Support: Acknowledgements: NIH EY001792, Dept of VA, RPB
342 Ocular angiography and oxymetry
Tuesday, May 05, 2015 11:00 AM–12:45 PM
Contributing Section(s): Clinical/Epidemiologic Research,
Physiology/Pharmacology, Retinal Cell Biology, Retina
Can OCT be accurate in quantifying retinal oxygen metabolism?
Hao F. Zhang, Ji Yi, Siyu Chen, wenzhong Liu. Biomedical
Engineering, Northwestern University, Evanston, IL.
Purpose: To explore whether OCT can be a potential candidate for
accurate measurement of retinal oxygen metabolism towards early
diagnosis of ischemic retinal diseases.
Methods: We first used statistical methods to numerically simulate
photon transport in the retina to mimic OCT working under different
spectral ranges. Then we analyze accuracy of OCT oximetry subject
to parameter variations such as vessel size, pigmentation, and
oxygenation. We then developed an experimental OCT system based
on the spectral range identified by our simulation work. We applied
the newly developed OCT to measure both retinal hemoglobin
oxygen saturation (sO2) and retinal retinal flow. To measure blood
flow, we performed double-circular-trajectory scans around the optic
disk to obtain the absolute blood velocity. After obtaining the retinal
sO2 and blood velocity, we further measured retinal vessel diameter
and calculated the retinal oxygen metabolism rate (MRO2). To test
the capability of our OCT, we imaged wild-type Long-Evans rats
ventilated with both normal air and air mixtures with various oxygen
concentrations.
Results: Our simulation suggested that OCT working within visible
spectral range is able to provide accurate measurement of retinal
MRO2 using invers Fourier spectral reconstruction. We refer our
newly developed OCT as vis-OCT, and showed that vis-OCT was
able to measure the sO2 value in every single major retinal vessel
around the optical disk as well as in micro retinal vessels as shown in
Figure 1. When breathing normal air, the averaged sO2 in arterial and
venous blood in Long-Evans rats was measured to be 95% and 72%,
respectively. When we challenge rats using air mixtures with different
oxygen concentrations, vis-OCT measurement follows analytical
models of retinal oxygen diffusion and pulse oximeter well.
Conclusions: Vis-OCT is a sensitive tool to measure retinal MRO2
with a high repeatability. It opens up a new window to investigate
several significant blinding diseases, such as diabetic retinopathy and
glaucoma, which strongly associate with retinal oxygen metabolic
disorders.
Commercial Relationships: Hao F. Zhang, Opticent Health (I); Ji
Yi, None; Siyu Chen, None; wenzhong Liu, None
Support: NIH 1R01EY019951, 1R24EY022883, and NSF CBET1055379
Error analysis of two-wavelength algorithms for retinal oximetry
J C. Ramella-Roman1, 2, Daniel Rodriguez2, Quanzeng Wang3, Joshua
Pfefer3. 1BME and Herbert Wertheim College of Medi, Florida
International University, Miami, FL; 2Biomedical Engieneering
Department, Florida International University, Miami, FL; 3Center
for Devices & Radiological Health, Food and Drug Administration,
Silver Spring,, MD.
Purpose: A new imaging technique for the assessment of retinal
vascular oximetry has been introduced recently. The technique
utilizes two-wavelength illumination of the fundus for the estimation
of oxygen saturation in the retina. Utilizing a well-calibrated, threedimensional stochastic model of light transport, we have examined
the error associated with the two-wavelength technique (570 nm, 600
nm). We analyzed the error produced when the analyzing algorithm
calibration assumptions are incorrect, vessel diameter varies,
choroidal melanin concentration varies, and when there is vascular
crosstalk from the choroid.
Methods: The voxel-based Monte Carlo code used is capable of
handling heterogeneous tissue structures, such as a vessel embedded
in a stratified medium with different optical properties. A simplified
tissue geometry comprised of four layers was used: retina, RPE,
choroid, and sclera. A vessel embedded within the retina was also
implemented. Two hundred million photons were launched in each
simulation. Photons back-reflected by the layers were used to create
images of fundus vasculature and then analyzed using the standard
two wavelength algorithm approach. This included the calibration
mechanism where a representative artery and a vein were selected
and their values were fixed to 96% and 54 % respectively.
Results: Under ideal conditions, two-wavelength retinal oximetry
estimates compare well with true oxygen saturation levels of the
vessel of interest. When the assumed values of calibrating arteries
or veins do not correspond to true values, errors as high as 20% in
veins and 10% in arteries may occur. When vessel size is increased,
the optical density ratio (ODR) decreases and the assessed oxygen
saturation error can be as high as 15% of its true value. Similar trends
were found when analyzing the effect of melanin in the choroid and
choroidal vessel crosstalk.
Conclusions: Two-wavelength retinal oximetry is a popular
technique because of its simplicity, however, it may be prone to
significant error. An advanced light-tissue interaction modeling
approach was effective in quantifying and elucidating key sources
of error. Although some have attempted to correct oxygen saturation
measurements for differences in pigmentation or vessel size, there
are still many confounding variables that may degrade accuracy; this
may lead to misdiagnosis of a range of retinal pathologies.
Commercial Relationships: J C. Ramella-Roman, None; Daniel
Rodriguez, None; Quanzeng Wang, None; Joshua Pfefer, None
a)Three dimensional view of retina angiography. b)En face view of
sO2 mapping. Scale: 200 mm.
Commercial Relationships: Siyu Chen, None; Ji Yi, None; Hao F.
Zhang, None
Support: NIH Grant 1R01EY019951, NIH Grant 1R24EY022883,
NSF Grant CBET-1055379, NSF Grant CBET-1066776, NSF Grant
DBI-1353952
Dynamically measuring retinal oxygen saturation at
microvascular level using visible-light OCT angiograph
Siyu Chen1, Ji Yi1, Hao F. Zhang1, 2. 1Department of Biomedical
Engineering, Northwestern University, Evanston, IL; 2Department of
Opthamology, Northwestern University, Chicago, IL.
Purpose: Although hemoglobin oxygen saturation (sO2) is a vital
physiological indicator for the functionality of retina tissues, accurate
measurement of ocular microvasculature sO2 remains challenging.
Here we propose to use optical coherence tomography based
angiography working within visible-light spectral range (Vis-OCT) to
measure absolute retina sO2.
Methods: We developed a Vis-OCT system working from 520 nm
to 630 nm to image rodent retina in vivo. Long Evans rats were
anesthetized using Ketamine/Xylazine cocktail. Once the animal
stabilized, it was ventilated normal air (21% O2) and a reference
image was taken. Then we supplied the following inhalation gas
mixtures in random order: (1) pure oxygen; (2) carbon dioxide
oxygen mixture (21% O2, 5% CO2); and (3) low oxygen air (10%
O2). For each inhalation gas, the animal was allowed to stabilize for 3
minutes, and normal air was supplied for at least 3 minutes between
changes. We took one additional measurement after the sequence
with normal air ventilation. During the experiment, a pulse oximeter
was attached to the hind limb to monitor the peripheral capillary
oxygen saturation (spO2). The acquired OCT images were processed
using phase-sensitive decorrelation algorithm. A series of short time
Fourier transform allowed the extraction of the sO2 encoded spectral
information. We calculated the first derivative of the wavelengthdependent angiography intensity (A’) around 574 nm. The value was
then calibrated into absolute sO2 percentage using spO2 readings.
Results: A linear regression showed strong positive correlation
between A’ and spO2 readings (R2 = 0.98). When inhaling normal
air, the sO2 was 80.6±7.4% for arteries and 65.2±4.9% for veins (n
= 8). During inhalation of gas 1 and 2, both artery and veins showed
significant increase in sO2 (A: p < 0.05; V: p < 0.01). The increase
was more dramatic in veins and when gas 1 was inhaled. We also
observed a significant decrease in sO2 when gas 3 was supplied (p <
0.01). When normal air was supplied again, sO2 return to its original
value for both arteries (p = 0.65) and veins (p=0.10).
Conclusions: Our results demonstrated that Vis-OCT angiography
can accurately measure retinal sO2 and its variations, which can
Fluorescein Angiography and Retinal Vessel Oxygen Saturation
in Patients with Proliferative Diabetic Retinopathy
Nicole K. Scripsema, Chavakij Bhoomibunchoo, Paul Whitten,
Robert Masini, Richard B. Rosen. Ophthalmology, The New York
Eye and Ear Infirmary of Mount Sinai, New York, NY.
Purpose: To determine oxygen saturation differences between areas
of active versus inactive Proliferative Diabetic Retinopathy (PDR)
using the Oxymap Retinal Oximeter.
Methods: Following a previous study which demonstrated increased
oxygen saturation in retinal vessels proportional to the severity of
Diabetic Retinopathy1, a retrospective review of diabetic patients
with PDR was performed. Patients imaged using both fluorescein
angiography (FA, Topcon, USA) and Oxymap T1 (Reykjavik,
Iceland) were included. Relative oxygen saturation of retinal vessels
was analyzed with the Oxymap. Retinal vessels associated with areas
of neovascularization on FA were compared to areas of inactive
disease. Paired t-tests were used to compare oxygen saturations.
Results: 48 eyes of 29 patients were included. 10 eyes were excluded
due to inadequate image quality. Mean age was 53.83±9.11 yrs. 62%
were female, 38% male. 44% were Hispanic, 26% African Americn,
11% Asian, 10% Caucasion, and 9% Other. Mean duration of diabetes
(DM) was 19.85±8.83 yrs. Mean hemoglobin A1c was 9.47±3.81%.
A majority of patients were initially diagnosed with Type II DM
(82.8%), now requiring both insulin and oral hypoglycemics
for glycemic control (84.6%). Mean relative arterial oxygen
saturation was higher in vessels associated with neovascularization
when compared to vessels in regions with no active disease
(107.93±13.85%, 104.87±14.25%, p=0.14). Mean relative venous
oxygen saturation was significantly lower in vessels associated with
neovascularization when compared to vessels in regions with no
active disease (66.05±11.87%, 69.57±9.19%, p=0.02).
Conclusions: These results suggest oxygen saturation in retinal
vessels associated with neovascularization varies from areas of
inactive disease. Our findings of higher arterial oxygen saturation
are consistent with a previous study that reported increased
oxygen saturation proportional to the severity of retinopathy.
The lower venous oxygen saturation in vessels associated with
neovascularization could be related increased leakage in areas of
angiogenesis. Initial data suggests the Oxymap could serve as a
clinically useful non-invasive imaging method for monitoring disease
be invaluable for the early diagnosis of several retinal metabolic
diseases.
progression. Further studies are needed to unravel the correlation
between Oxymap results and FA findings.
Reference:
1. Jørgensen CM et al. Acta Ophthalmol. 2014 Feb;92(1):34-9.
Commercial Relationships: Nicole K. Scripsema, None; Chavakij
Bhoomibunchoo, None; Paul Whitten, None; Robert Masini,
None; Richard B. Rosen, Advanced Cellulat Technologies (C),
Allergan (C), Carl Zeiss Meditech (C), Clarity (C), OD-OS (C),
Opticology (I), Optovue (C)
Support: The Bendheim-Lowenstein Retina Center Fund of the New
York Eye and Ear Infirmary
Oxygen Metabolism of the Inner Retina in the 50/10 Rat Model of
Retinopathy of Prematurity
Brian Soetikno1, 2, Ji Yi2, Patryk Purta1, wenzhong Liu2, Ronil
S. Shah1, Hao F. Zhang2, Amani A. Fawzi1. 1Department of
Ophthalmology, Northwestern University Feinberg School of
Medicine, Chicago, IL; 2Department of Biomedical Engineering,
Northwestern University, Functional Optical Imaging Laboratory,
Evanston, IL.
Purpose: To compare the inner retinal metabolic rate of oxygen
(rMRO2) in normal rats and rats with 50/10 oxygen-induced
retinopathy (OIR) using visible-light optical coherence tomography
(vis-OCT) at P18.
Methods: Beginning at birth, four Sprague-Dawley rat pups were
exposed to alternating hyperoxia (50% O2) and hypoxia (10% O2)
every 24 hours for 14 days (OIR group), while six rat pups were
allowed to grow at room air (control group). On P14, the OIR
pups were returned to room air. On P18, imaging was performed
to measure the oxygen saturation of hemoglobin (sO2) of the inner
retinal arterioles and venules for both control and OIR groups [A,
B]. These experiments used a vis-OCT system, which incorporated
a supercontinuum source with wavelengths from 500 nm to 620
nm. A dual-circle scanning protocol implemented on the same OCT
system enabled the measurement of vessel diameter, blood velocity,
and volumetric blood flow [C, D]. Using the sO2 measurements,
we calculated the oxygen extraction fraction (OEF) of the inner
retina. Combining the OEF with the total blood flow permitted the
calculation of the rMRO2. Following the imaging experiments on
P18, the retinas were harvested, immunostained with Alexa Fluor
594 isolectin, and imaged using fluorescence microscopy [E]. We
quantified the percentage of vaso-obliteration and counted the
number of clock-hours with neovascularization in the OIR retinas.
An unpaired Student’s t-test was used to compare the measurements
between control and OIR groups.
Results: There was no significant difference in the OEF between the
control and OIR groups (0.231 ± 0.037 vs. 0.262 ± 0.054; p=0.3550).
The average total estimated blood flow was significantly lower in
OIR versus control groups (2.74 ± 0.58 μl/min vs. 7.37 ± 2.96 μl/
min; p=0.003). Additionally, rMRO2 was significantly decreased
in OIR versus control groups (128 ± 32 nl min-1 vs. 338 ± 138 nl
min-1; p = 0.0108). For the OIR group, the average vasoobliteration
was 10.92 ± 3.50% and the average number of clock hours with
neovascularization was 4.25 ± 2.25 clock hours.
Conclusions: We observed a 59% reduction in the rMRO2 in the
OIR group as compared with that of control group, suggesting that
the metabolism of the inner retina is markedly reduced at P18 in the
50/10 OIR rat model.
Commercial Relationships: Brian Soetikno, None; Ji Yi, None;
Patryk Purta, None; wenzhong Liu, None; Ronil S. Shah, None;
Hao F. Zhang, None; Amani A. Fawzi, None
Support: R01EY019951 (AAF, HFZ) and R21HD077336 (AAF),
Research to Prevent Blindness, NY (Department of Ophthalmology,
Northwestern University)
Oxygen saturation profiles in Asian Indian eyes with central
retinal artery occlusion (CRAO)
Ashwin Mohan1, Priya Srinivasan1, Supriya Dabir1, Rajani Battu1,
Naresh K. Yadav1, Rohit Shetty2. 1Retina, Narayana Nethralaya,
Bangalore, India; 2Vice-Chairman, Narayana Nethralaya, Bangalore,
India.
Purpose: To study the oxygen saturation profiles in Asian Indian eyes
with central retinal artery occlusion (CRAO)
Methods: This is a retrospective review of 10 patients presenting to
our hospital in the last 6 months with a history of sudden painless
loss of vision and diagnosed as CRAO on clinical examination. In
addition to a comprehensive ophthalmic examination, they underwent
retinal oximetry (Oxymap T1, Oxymap hf, Reykjavik, Iceland). 2
patients had oximetry pre and post anterior chamber paracentesis and
3 patients had a follow up Oximetry at monthly intervals. Optic disc
centered images taken after dilatation were analysed by selecting the
thickest arteriole and venule per quadrant for saturation and diameter.
Results: The oxygen saturation was decreased in both the arterioles
(86%) and venules (50%). The arteriolar and venous diameters and
the arterio-venous saturation difference did not vary significantly
between the groups. In the 3 patients on follow up we saw an
improvement of arteriolar saturations in all and venous values in 2
of them. Two patients who underwent anterior chamber paracentesis
showed an improvement of both arteriolar and venous saturation
values after paracentesis.
Conclusions: Oximetry in CRAO shows an initial fall in saturations
with an increase over time or after paracentesis.
venous diameter (Pearson correlation coefficient = -0.513, P=0.017).
The relative partition of arterial and venous retinal oxygen saturation
to the cerebral oxygen saturation value was 31.5% and 68.5%
respectively.
Peripheral oxygen saturation was correlated neither with cerebral nor
with retinal oxygen saturation.
Conclusions: This is the first study to show a correlation between
retinal and cerebral oximetry, as measured by non-invasive
spectroscopy.
The observed relative partition of arterial and venous retinal oxygen
saturation is very close to the established calibration of FORESIGHT NIRS technology to a weighted average of 30% arterial and
70% venous contribution. Therefore, our first findings support a
promising role for retinal oximetry as a non-invasive reliable tool in
evaluating cerebral oxygenation in health and disease.
Commercial Relationships: Karel Van Keer, None; Cathy De
Deyne, None; Cornelia Genbrugge, None; Luis Pinto, None;
Evelien Vandewalle, None; Ingeborg Stalmans, None
Figure a & b of Patient 1 and c & d of Patient 2 at presentation and
immediately after paracentesis respectively
Commercial Relationships: Ashwin Mohan, None; Priya
Srinivasan, None; Supriya Dabir, None; Rajani Battu, None;
Naresh K. Yadav, None; Rohit Shetty, None
Retinal Blood Oxygen Saturation and Aqueous Humor
Biomarkers in Early Diabetic Retinopathy
Faryan Tayyari1, Lee-Anne Khuu2, Jeremy M. Sivak2, Shaun Singer2,
Michael H. Brent2, John G. Flanagan1, 2, Christopher Hudson1, 2.
1
Optometry and Vision Science, University of Waterloo, Waterloo,
ON, Canada; 2Ophthalmology and Vision Sciences, University of
Toronto, Toronto, ON, Canada.
Purpose: To investigate the relationship between retinal blood
oxygen saturation (SO2) and aqueous humor (AH) concentrations of
protein biomarkers in diabetic patients with non-proliferative diabetic
retinopathy (NPDR) and to compare them with those of control
subjects.
Methods: The sample comprised 14 participants with mild-tomoderate NPDR (66.3 ± 9.1 years) and 17 age-matched controls
(69.1 ± 5.7 years); all participants were previously scheduled for
routine cataract extraction with intraocular lens implantation. At the
start of surgery, AH was collected using a Sautter hydrodissection
cannula (27G, 0.4x22mm; Geuder, Heidelberg, Germany). Multiplex
cytokine analyses of 26 biomarkers, including Angiopoietin 2 (Ang
2), Epidermal Growth Factor (EGF), Hepatocyte Growth Factor
(HGF) and Interleukin-8 (IL-8) were performed by BioPlex 200
system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Four
weeks after uncomplicated surgery and 2 weeks after cessation of any
anti-inflammatory eyedrops, non-invasive hyperspectral retinal (HR)
imaging (prototype H-8.5 HR Camera, Optina, QC, Canada) was
undertaken. Six repeated retinal images were acquired centered on
the optic nerve head at wavelengths of 586 and 605nm using the HR
camera for each subject.
Results: Mean venular retinal blood oxygen saturation (p<0.001) and
AH levels of HGF (p=0.018), Ang 2 (p=0.005), EGF (p=0.030) and
IL-8 (p=0.034) were significantly higher in NDPR when compared
to controls. This study demonstrated a correlation between venular
retinal blood oxygen saturation and pro-angiogenic factors HGF
(r=0.584, p=0.030), Ang 2 (r=0.592, p=0.026), and EGF (r=0.523,
p=0.050), but did not find any correlation for IL-8 (r=0.435, p=0.120)
even though this biomarker was significantly higher in the diabetic
group.
Conclusions: The results of this study revealed a direct relationship
between retinal blood SO2 and Ang 2, EGF and HGF but not IL-8.
Commercial Relationships: Faryan Tayyari, None; Lee-Anne
Khuu, None; Jeremy M. Sivak, None; Shaun Singer, None;
Retinal oximetry as a novel parameter in the non-invasive
assessment of cerebral oxygenation
Karel Van Keer1, Cathy De Deyne2, 3, Cornelia Genbrugge2, 3, Luis
Pinto4, Evelien Vandewalle1, Ingeborg Stalmans1, 5. 1Department of
Ophthalmology, University Hospitals Leuven, Leuven, Belgium;
2
Department of Anesthesiology, Intensive Care, Emergency Medicine
and Pain Therapy, Ziekenhuis Oost-Limburg ZOL, Genk, Belgium;
3
Faculty of Medicine and Life Sciences, Biomedical Research
Institute, Hasselt University, Diepenbeek, Belgium; 4Department of
Ophthalmology, Faculty of Medicine of Lisbon, Lisbon, Portugal;
5
Department of Ophthalmology Neurosciences, Laboratory of
Ophthalmology, KU Leuven, Leuven, Belgium.
Purpose: To investigate the relation between retinal and cerebral
oximetry, as measured by non-invasive spectroscopy and to establish
the relative partition of arterial and venous retinal oxygen saturation
to the measured cerebral oxygen saturation.
Methods: In this observational, cross-sectional study we performed
peripheral, retinal and cerebral oxygen saturation measurements in
healthy volunteers during steady state conditions. Peripheral oxygen
saturation was measured on the left index finger by pulse oximetry
(Ohmeda TruSat Pulse Oximeter, GE Healthcare, Finland). Arterial
and venous retinal oxygen saturation and vessel diameter were
measured in the left eye using a noninvasive spectrophotometric
retinal oximeter (Oxymap T1, Oxymap ehf., Reykjavik, Iceland).
Cerebral oxygen saturation was measured over the left cerebral
cortex using cerebral near-infrared spectroscopy (NIRS) (FORESIGHT technology, CAS Medical Systems, Branford, CT).
Correlations between the parameters were assessed using Pearson
correlation coefficients. The relative arterial and venous partition was
determined.
Results: 21 young healthy individuals aged 26.4±2.2 years were
analyzed. Cerebral oxygen saturation values showed a significant
positive correlation with both arterial and venous retinal vessel
oxygen saturation values (r = 0.442, P=0.045 and r = 0.434 P=0.049
respectively) and a significant negative correlation with retinal
Michael H. Brent, None; John G. Flanagan, None; Christopher
Hudson, Optina (F)
Support: Ontario Research Fund for Research Excellence
Computer-based quantification of acellular capillaries to assess
experimental diabetic retinopathy
Craig Schebler1, 2, Mihran Tuceryan1, Jiang-Yu Zheng1, Ashay D.
Bhatwadekar2. 1Department of Computer and Information Science,
Indiana University Purdue University Indianapolis, Indianapolis, IN;
2
Department of Ophthalmology, Indiana University, Indianapolis, IN.
Purpose: Increase in acellular capillary numbers is the pathologic
hallmark of diabetic retinopathy. Traditionally, the acellular
capillaries are enumerated by blinded investigators either directly
through a microscope or via manual counting of captured images.
However, this system is laborious, time consuming, and often shows
inconsistency amongst researchers. The purpose of this study is
to create a computer based algorithm that will assess the acellular
capillaries of the retina, consistently reducing the human error and
time.
Methods: The retinas of control and diabetic mice were processed
using trypsin digestion and the high resolution .tiff images of
retinal quadrants. We used the Python programming language using
assorted open source package to write a custom-designed code. The
images underwent a Gaussian blur and noise reduction to clean up
the imperfections of the image. We used a purpose-built k-means
clustering algorithm to group similar parts of the image together.
We generated the paths in each image by converting all non-white
elements to black. These images were then processed for Medial
Access Transform (MAT) to create the skeleton as well as to find the
distance from the skeleton to the edges formed in the above steps.
The colors represent the distance from the edges, in which red is the
largest distance and purple is the shortest distance. Then the locations
where the skeleton is purple and connects to another color on both
ends were counted. This count is the number of acellular capillaries.
Results: We have developed a precise algorithm with improved
accuracy to enumerate the numbers of acellular capillaries. This
algorithm can be used to quickly count the acelluar capillaries in
diabetic retinas and to create a standard for retinopathy assessment.
Moreover, this algorithm is compatible with open source image
analysis programs, enabling ease of access to the users.
Conclusions: We have designed an automated computer-based
system to enumerate the acellular capillaries in diabetic retina. This
computer based automated system will enhance consistency in
retinopathy assessment and reduce time for analysis.
Figure 1 Schematics of serial image processing as described in
methods.
Figure 2 Identification of acellular capillaries (blue arrow & purple
area in inset) after MAT.
Commercial Relationships: Craig Schebler, None; Mihran
Tuceryan, None; Jiang-Yu Zheng, None; Ashay D. Bhatwadekar,
None
Support: LHSI Internship and Glick Eye Institute Foundation
The Association of Retinal Blood Flow and Retinal Blood Oxygen
Saturation in Mild-to-Moderate Diabetic Retinopathy
Christopher Hudson1, 2, Faryan Tayyari1, Lee-Anne Khuu2,
Shaun Singer2, John G. Flanagan1, 2, Michael H. Brent2. 1School
of Optometry, University of Waterloo, Waterloo, ON, Canada;
2
Ophthalmology and Vision Sciences, University of Toronto, Toronto,
ON, Canada.
Purpose: The aim of this study was to evaluate the relationship
between retinal blood flow (RBF) and retinal oxygen saturation (SO2)
in mild-to-moderate non-proliferative diabetic retinopathy (NPDR)
and in age-matched controls.
Methods: One eye of each of 15 healthy subjects (68.1 ± 6.0
years) and 13 subjects with mild-to-moderate NPDR (67.3 ± 10.2
years) was dilated. None of the patients with NPDR had received
treatment for their retinopathic changes or had any evidence of
sight-threatening characteristics. Doppler FD-OCT blood flow was
measured using the prototype RTVue system (Optovue Inc., USA);
six separate measurements each comprising an upper and a lower,
nasal pupil scan were acquired. Non-invasive hyperspectral retinal
(HR) imaging (prototype H-8.5 HR Camera, Optina, QC, Canada)
was also undertaken to measure retinal blood oxygen saturation; six
HR camera (including the 586 to 605nm wavelengths) measurements
were acquired.
Results: Total retinal blood flow was significantly lower in NDPR
when compared to controls (42.66 ± 7.5 vs 33.73 ± 9.13; p=0.004).
Mean retinal venular blood oxygen saturation was higher in NDPR
than in the healthy controls (62.55 ± 5.7% vs 56.29 ± 4.7%; p=0.003).
However, there was no correlation between venular retinal blood flow
and venular oxygen saturation in controls (r=0.243, p=0.34) or in DR
(r=0.228, p=0.45).
Conclusions: There was no correlation between venular retinal blood
flow and retinal blood oxygen saturation. Given that there was no
correlation between SO2 and RBF, the results of this study suggest
that there is a need to measure both SO2 and RBF in order to calculate
retinal oxygen utilization.
Commercial Relationships: Christopher Hudson, Optina (F),
Optovue Inc (F); Faryan Tayyari, None; Lee-Anne Khuu, None;
Shaun Singer, None; John G. Flanagan, None; Michael H. Brent,
None
Support: Ontario Research Fund for Research Excellence
Image Quality Affects Measurements of Retinal Vessel Oxygen
Saturation
Sveinn H. Hardarson1, Benedikt A. Jonsson2, Robert A. Karlsson3,
Asbjorg Geirsdottir4, David Bragason1, Thor Eysteinsson1, 5,
Olof B. Olafsdottir1, Jona V. Kristjansdottir1, Einar Stefansson1.
1
Ophthalmology, University of Iceland / Landspítali University
Hospital, Reykjavik, Iceland; 2Electrical- and Computer Engineering,
University of Iceland, Reykjavik, Iceland; 3Oxymap ehf., Reykjavik,
Iceland; 4St. Eriks Eye Hospital, Stockholm, Sweden; 5Physiology,
University of Iceland, Reykjavik, Iceland.
Purpose: Non-invasive measurements of oxygen saturation in retinal
vessels are based on retinal images, taken at two wavelengths of light.
The purpose of the study was to test if and how image quality affects
measurements of retinal vessel oxygen saturation.
Methods: The retinal oximeter (Oxymap ehf., Iceland)
simultaneously acquires images of the retina at 570nm and 600nm.
Software automatically measures light absorbance of retinal vessels
and calculates oxygen saturation. A newly developed software tool
automatically grades the images on the scale of 0 to 1 according to
quality of the images. The quality grade is composed of assessment
of focus and contrast. Oximetry images of 108 healthy individuals
(age 18-77) and 17 individuals, scheduled for cataract surgery were
analyzed for quality.
Results: In the heatlhy individuals, measured venous oxygen
saturation decreased with worsening image quality (p<0.0001,
R2=0.17) but measured arteriolar oxygen saturation was not
significantly affected (p=0.79). A linear model indicated
that the relationship between quality and measured venous
saturation was not due to age: VenousSaturation=260.043*Age+37*ImageQualityGrade, p=0.29 for age and p=0.0008
for image quality. Further analysis showed that the effect of image
quality was due to contrast (p=0.0002) rather than focus (p=0.36). In
individuals scheduled for cataract surgery, image quality correlated
positively with measured oxygen saturation in both arterioles
(p=0.0079, R2=0.38) and venules (p=0.0034, R2=0.45).
Conclusions: Poor image quality leads to lower measured
venous oxygen saturation and, in more extreme cases, also affects
measurements of arterioles. This may explain why measured oxygen
saturation has previously been found to decrease with age. The
new tool for image quality assessment allows quality control of
retinal oximetry measurements and may potentially be used to apply
correction factors.
The figure shows that measured retinal venous oxygen saturation
increases with improved image quality as assessed by an automatic
image quality tool. The results of simple linear regression are shown
(not corrected for age).
Commercial Relationships: Sveinn H. Hardarson, Oxymap ehf.
(C), Oxymap ehf. (I), Oxymap ehf. (P); Benedikt A. Jonsson,
Oxymap ehf. (F); Robert A. Karlsson, Oxymap ehf. (E), Oxymap
ehf. (I), Oxymap ehf. (P); Asbjorg Geirsdottir, None; David
Bragason, None; Thor Eysteinsson, Oxymap ehf. (I), Oxymap ehf.
(P); Olof B. Olafsdottir, None; Jona V. Kristjansdottir, None;
Einar Stefansson, Oxymap ehf. (I), Oxymap ehf. (P), Oxymap ehf.
(S)
Support: The Icelandic Center for Research (Rannís), University
of Iceland Research Funds, The Landspítali University Hospital
Research Fund, Helga Jónsdóttir and Sigurliði Kristjánsson Memorial
Fund
Validation of model based hyperspectral retinal oximetry
algorithms using systemic gas provocations in healthy individuals
Susith Kulasekara1, Kalpana Rose2, Michèle Desjardins4, Reza
Jafari3, J Daniel Arbour5, 3, Frédéric Lesage4, Jean-Philippe
Sylvestre3, Christopher Hudson1, 2. 1Ophthalmology & Vision
Sciences, University of Toronto, Toronto, ON, Canada; 2School of
Optometry and Vision Science, University of Waterloo, Waterloo,
ON, Canada; 3Optina Diagnostics, Montreal, QC, Canada; 4Institut
de génie biomédical, École Polytechnique de Montréal, Montréal,
QC, Canada; 5Department of Ophthalmology, University of Montreal,
Montreal, QC, Canada.
Purpose: Retina is a highly metabolically active tissue with a very
high demand for oxygen. Dysregulation of retinal oxygen supply
and demand is associated with many ocular and systemic diseases.
Changes in retinal tissue oxygen tension may take place before these
changes are reflected in retinal vessels. The purpose of this study is
to validate the use of a model based on a hyperspectral algorithm for
measuring retinal tissue oxygen saturation (tSO2).
Methods: One eye of 12 healthy non-smoking volunteers was chosen
for the study. End-tidal O2 concentration (PETO2) was adjusted using
a model based prospective targeting device (RespirAct) to induce
normoxia (PETO2=100mmHg), hypoxia (PETO2=50mmHg) and
hyperoxia (PETO2=300mmHg), while maintaining normocarbia. The
order of hyperoxia and hypoxia was randomized between subjects.
Heart rate, blood pressure, and finger pulse oximetry were monitored
throughout. A prototype metabolic hyperspectral retinal camera
(MHRC, Optina Diagnostics) was used to image the fundus from
500-600nm in 5nm steps (3 repeats per condition) after stabilization
of finger pulse oximetry for over 3 min. The reflectance intensity data
was fit in MATLAB to a model where oxy- and deoxyhemoglobin are
the main absorbers and scattering is modeled by a log(1/ wavelength)
term. The fitted parameters were used to extract an estimation of tSO2
and total hemoglobin content (HbT) in each pixel of the images and
values obtained in the different PETO2 conditions were compared for
a region of the retina, free of any visible blood vessels, at a half disc
diameter distance from the disc margin.
Results: The preliminary results show that as the breathing air
PETO2 was increased from normoxia to hyperoxia tSO2 significantly
increased (p=0.001) from 41%(+11) to 53%(+10). Lowering the
PETO2 from normoxia to hypoxia significantly decreased (p=0.001)
tSO2 from 41%(+11) to 34%(+14). The mean HbT at hypoxia,
normoxia, and hyperoxia, were not significantly different (p=0.3)
from each other: 2.8(+0.7); 2.5(+0.6); 2.4(+0.8). However, there was
a trend towards an increase in HbT in hypoxia.
Conclusions: As the breathing air oxygen composition ( PETO2) is
changed from normoxia to hypoxia and hyperoxia, retinal tissue
oxygen saturation (tSO2) measurements, based on a hyperspectral
algorithm, showed parallel changes, suggesting that the method could
be used to monitor the health of the retina.
Commercial Relationships: Susith Kulasekara, None; Kalpana
Rose, None; Michèle Desjardins, None; Reza Jafari, Optina
Diagnostics (E); J Daniel Arbour, Optina Diagnostics (I); Frédéric
Lesage, None; Jean-Philippe Sylvestre, Optina Diagnostics (E),
Optina Diagnostics (I); Christopher Hudson, Optina Diagnostics
(F), Thornhill Research Inc. (P)
Support: Vision Science Research Program (VSRP) Award,
Department of Ophthalmology, University of Toronto, Canada;
Ontario Research Fund – Research Excellence (ORF-RE); Dalton
Whitebread Scholarship Fund,Faculty of Medicine, University of
Toronto.
OCT Angiography (OCTA) in Healthy Human Subjects
Jack Yi1, Douglas Matsunaga1, John E. Legarreta2, Andrew D.
Legarreta2, Giovanni Gregori2, Mary K. Durbin3, Utkarsh Sharma3,
Philip J. Rosenfeld2, Carmen A. Puliafito1, Amir H. Kashani1. 1USC
Eye Institute, Keck School of Medicine of USC, Los Angeles, CA;
2
Bascom Palmer Eye Institute, University of Miami Miller School of
Medicine, Miami, FL; 3Advanced Development, Carl Zeiss Meditec,
Inc., Dublin, CA.
Purpose: To evaluate the feasibility of noninvasive retinal
angiography using a prototype using swept-source (SS) and spectraldomain (SD) optical coherence tomography (OCT) angiography.
Methods: Data was acquired using a Cirrus (Carl Zeiss Meditec,
Dublin, CA) SS-OCT and SD-OCT prototype angiography systems.
Five healthy subjects (nine eyes) with no past ophthalmologic history
were recruited. 3x3mm regions centered on the fovea, nasal macula,
and temporal macula were imaged. Retinal vasculature was assessed
in three horizontal slabs consisting of the inner, middle, and outer
retina. The vasculature was reconstructed using an intensity-based
algorithm into separate en face images. Post-processed en face OCT
angiograms were analyzed with ImageJ (NIH, Bethesda, MD) to
quantify the density of retinal microvasculature using the “Auto
Local Threshold” plug-in (Landini G. v1.5).
Results: OCTA in healthy subjects resembled fine capillary networks
that have been demonstrated in previous histological studies of
human cadaver eyes. Retinal vessels were not visualized in the
outer retina. Within the central macula and temporal macula, the
inner retina displayed continuous capillaries traveling in the same
retinal plane while the middle retinal slab contained a lattice pattern
of vessels. The nasal macular region showed capillary segments
radiating out from the optic disc in the inner retina while the
middle retinal slab featured a lattice pattern of discontinuous vessel
segments. Vessel density analyses using ImageJ are shown in Figures
1 and 2.
Conclusions: Noninvasive, high-resolution angiograms produced
by OCTA show qualitatively similar vascular patterns to previous
histological images of the retina. OCTA angiography can reliably
and reproducibly image the fine capillary networks of the retina
and may have a role in assessing the retinal microvasculature when
conventional fluorescein angiography cannot be performed.
Fig. 1. Vessel density analysis of the central macular region of a
healthy subject yielded an average total density of 31.68% ± 1.15% in
the inner retina and a density of 30.86% ± 1.20% in the middle retina.
Fig. 2. Vessel density analysis of the nasal macular region of a
healthy subject yielded an average total density of 31.59% ± 1.40% in
the inner retina and a density of 31.47% ± 1.62% in the middle retina.
Commercial Relationships: Jack Yi, Carl Zeiss Meditec (F);
Douglas Matsunaga, Carl Zeiss Meditec (F); John E. Legarreta,
Carl Zeiss Meditec (F); Andrew D. Legarreta, Carl Zeiss Meditec
(F); Giovanni Gregori, Carl Zeiss Meditec (F); Mary K. Durbin,
Carl Zeiss Meditec (E); Utkarsh Sharma, Carl Zeiss Meditec (E);
Philip J. Rosenfeld, Carl Zeiss Meditec (F); Carmen A. Puliafito,
Carl Zeiss Meditec (F); Amir H. Kashani, Carl Zeiss Meditec (F)
Support: An Unrestricted grant from Research to Prevent Blindness,
New York, NY 10022
Comparison of Swept Source Versus Spectral Domain Optical
Coherence Tomography Angiography (OCTA) in Chorioretinal
Diseases
Lauren Branchini1, Talisa de Carlo1, Eric Moult2, Nadia K.
Waheed1, Andre Witkin1, Caroline R. Baumal1, James G. Fujimoto2,
Jay S. Duker1. 1Tufts New England Eye Center, Boston, MA;
2
Massachusetts Institute of Technology, Cambridge, MA.
Purpose: Optical coherence tomography angiography (OCTA) to
visualize chorioretinal vasculature can be performed using both 840
nm wavelength spectral domain (SD) as well as 1060 nm swept
source (SS) systems. This cross sectional, observational case series
compares and contrasts the imaging capabilities of SS-OCTA to SDOCTA in a variety of cases.
Methods: 9 subjects underwent sequential same-day SD-OCTA
and SS-OCTA at the New England Eye Center between 8/2014 and
11/2014. These subjects included 1 normal control and 8 with a
variety of chorioretinal diseases including wet age-related macular
degeneration, central serous chorioretinopathy, central retinal vein
occlusion, diabetic retinopathy, birdshot chorioretinopathy, Best’s
disease and retinitis pigmentosa. All subjects were imaged on a
prototype SS-OCTA system featuring an imaging speed of 400 kHz,
with an axial resolution of 7-10 um and 5 repeated B-scans per
position. All subjects were also imaged with the AngioVue OCTA
system on the commercially available Avanti SD-OCT (Optovue, Inc,
Fremont, CA) operating at 70 kHz with an axial resolution of 5um
using 2 repeated B-scans. An investigational OCTA software program
was used to evaluate the images. Images were compared qualitatively.
When available, OCTA was also compared with corresponding
fluorescein angiography (FA).
Results: High quality images were obtained on both systems in all
subjects. All data sets were segmented and vascular layers in the
retina and choroid were identified. Pathologic features of disease
were identified in images from both systems. These pathologic
features, including choroidal neovascular membranes (Figure 1),
retinal microaneurysms and vessel dropout were correlated between
scanning systems and with FA. Better quality images of the choroidal
vasculature were obtained with the SS-OCTA especially in patients
with ocular opacities including subretinal lipofuscin deposits. No
retinal vascular pathology could not be identified on both systems.
Conclusions: Both SS-OCTA and SD-OCTA have a role in
evaluating chorioretinal vasculature and demonstrate consistent
findings.
Commercial Relationships: Lauren Branchini, None; Talisa
de Carlo, None; Eric Moult, None; Nadia K. Waheed, Iconic
Therapeutics (C); Andre Witkin, None; Caroline R. Baumal, None;
James G. Fujimoto, Carl Zeiss Meditec, Inc. (P), Optovue (C);
Jay S. Duker, Carl Zeiss Meditec Inc. (P), EyeNetra (I), Hemera
Biosciences (I), Ophthotech Corp. (I)
Support: This work was supported in part by a Research to Prevent
Blindness Unrestricted grant to the New England Eye Center/
Department of Ophthalmology, Tufts University School of Medicine,
National Institutes of Health (NIH) contracts R01-EY011289-28,
R01-EY013178-12, R01-CA075289-16, Air Force Office of
Scientific Research contracts FA9550-10-1-0551 and FA9550-121-0499, DFG contracts DFG-HO-1791/11-1, DFG Training Group
1773, DFG-GSC80-SAOT, and Massachusetts Lions Club.
Comparison of two free retinal vascular measurement software
packages: IVAN and VAMPIRE
Elaine Downie1, Julian Tokarev1, Afshin Divani3, Dara D.
Koozekanani2. 1University of Minnesota Medical School,
Minneapolis, MN; 2Department of Ophthalmology, University of
Minnesota, Minneapolis, MN; 3Department of Neurology, University
of Minnesota, Minneapolis, MN.
Purpose: Retinal image analysis can be used to quantify retinal
vascular changes due to hypertensive retinopathy, changes shown
to be predictive of stroke risk. A number of software packages are
available for this, and each utilizes different techniques for detection
and measurement of retinal vascular features. Here, we compared the
performance of a well known package, IVAN, with a newly available
tool, VAMPIRE.
Methods: 50 degree color fundus images obtained with a Topcon
TRC 50DX camera (Topcon Medical Systems, Oakland, NJ) were
graded using IVAN (Interactive Vessel ANalyzer), v1.3 (courtesy of
Dr. Nicola Ferrier of the UW Madison School of Engineering and
the Dept. of Ophthalmology and Visual Sciences) and VAMPIRE
(Vascular Assessment and Measurement Platform for Images of the
Retina, available at vampire.computing.dundee.ac.uk). Both are
available free of charge. Both were used to calculate central retinal
artery equivalents (CRAE), central retinal vein equivalents (CRVE),
and arteriolar:venular ratio (AVR) for each eye.
Results: 21 patients, 42 eyes, were analyzed. CRAE, CRVE, and
AVR values were obtained for 41 eyes, one eye was ungradeable. The
values were plotted and a correlation coefficient between the IVAN
and VAMPIRE values was calculated and tested for significance
using the Pearson Product-Moment Significance Test. For CRVE, the
R2=0.7032, the Pearson correlation coefficient (R)=0.839, p<0.001.
For CRAE, the R2=0.2604, the Pearson correlation coefficient
(R)=0.510, p=0.001. For AVR, the R2=0.5747, the Pearson correlation
coefficient (R)=0.758, p<0.001.
Conclusions: The CRAE, CRVE, and AVR values measured with the
newer VAMPIRE software package corresponded with those obtained
using IVAN. While the two packages use different methods and have
different degrees of automaticity, this suggests both can be valuable
tools in the investigation of hypertensive retinopathy and other retinal
vascular conditions for which CRAE, CRVE, and AVR are relevant.
Figure 1. IVAN segments vessels within a ring between 0.5 and 1
disc diameters from the disc center. The grader verifies correct vessel
identification and revises the segmentations. CRAE, CRVE, and AVR
are calculated using the Parr-Hubbard-Knudtson equation.
Figure. 2A VAMPIRE displays concentric rings and allows the user
to designate points for vascular measurement.
Figure. 2B Vascular measurement is made manually with a
measuring tool.
Commercial Relationships: Elaine Downie, None; Julian Tokarev,
None; Afshin Divani, None; Dara D. Koozekanani, None
An improved semi-automatic method to measure retinal vessel
caliber from fundus photographs and measurement results in
normal Japanese
Makoto Araie1, Aiko Iwase2, Ryo Kawasaki3, Jun Suehiro4, Akihiko
Sekine4. 1Kanto Central Hospitals, Mutual Aid Assoc of Public Sch
Teachers, Setagaya-Ku, Japan; 2Ophthalmology, Tajimi Iwase Eye
Clinic, Tajimi, Japan; 3Yamagata Univ, Yamagata, Japan; 4Topcon,
Tokyo, Japan.
Purpose: Retinal vessel caliber (RVC) changes are associated with
various systemic and ocular abnormalities. Accurate magnification
correction in each eye is indispensable for measuring RVC from
fundus photographs, but currently available semi-automatic methods
to determine RVD are not necessarily satisfactory in this respect.
Methods: In addition to Littmann’s correction, ray-tracing algorithm
was applied to optical system of Topcon non-mydriatic fundus
camera (TRC-NW7, Topcon, Tokyo) and magnification was corrected
using corneal curvature, axial length (AL) and refraction of each
subject eye. RVC was always determined in an annular area between
1.8 and 2.7 mm from the disc center. Measurement reproducibility
was checked in 20 normal eyes by 4 independent examiners, and
the results with the current method were compared to those with
Interactive Vessel Analysis System (IVAN) in other 180 normal eyes
of 99 subjects with mean age of 52 yrs. Factors relating to central
retinal arteriolar equivalent (CRAE) were studied in 793 right eyes
with AL less than 26 mm of 793 normal Kumejima Study participants
with average age of 50 yrs. The Ethic Committee of Kumejima Town
approved the study.
Results: Intra-individual ICC was 0.921-0.993 and 0.995-0.999, and
Inter-individual ICC 0.998 and 0.996 for CRAE and central retinal
venular equivalent (CRVE), respectively. The current method yielded
CRAE and CRVE values 1.8 % smaller and .4.3% greater than those
with IVAN (P<0.017), and the difference between the results with
the two methods was AL-dependent. In normal Japanes, CRAE
was 162.5 +/- 9.3 (SD) and CRVE 232.9 +/- 16.1 um. A multiple
regression analysis revealed that female gender (P=0.002), older age
(P=0.000), higher blood pressure (P=0.014), longer AL (P=0.000)
and larger beta-peripapillary atrophy (beta-PPA) (P=0.000) were
correlated with CRAE in this population.
Conclusions: An improved method to measure RVC from fundus
photographs with reasonable measurement reproducibility was
developed. In normal Japanese, two ocular factors, axial length and
beta-PPA area, were newly found to significantly correlate with
CRAE beside systemic factors well known to affect it.
Total retinal blood flow measurement in the human eye with
3-Beam Doppler Optical Coherence Tomography
Richard Haindl1, 2, Wolfgang Trasischker1, 2, Bernhard Baumann1, 2,
Andreas Wartak1, 2, Michael Pircher1, 2, Christoph K. Hitzenberger1,
2 1
. Center for Medical Physics and Biomedical Engineering, Medical
University of Vienna, Vienna, Austria; 2Medical Imaging Cluster,
Medical University of Vienna, Vienna, Austria.
Purpose: To measure the total retinal blood flow and obtain flow/
velocity profiles within all major retinal vessels originating from the
optic nerve head (ONH) using an improved three-beam Doppler OCT
(D-)OCT technique.
Methods: The three beam D-OCT consists of three independent
superluminescent diode (SLD) sources with a central wavelength of
840 nm and a spectral bandwidth of 50 nm. The collimated exiting
beams share a common bulk optics Michelson interferometer. A welldefined beam geometry enables the full reconstruction of the three
dimensional velocity vector, without prior knowledge on the vessel
geometry, which is normally required for D-OCT systems with less
than three beams.
In the sample arm a custom made facet prism telescope allows for
variable beam separation adjustment without alteration of the beam
diameter. In conjunction with a normal telescope a transversal
resolution of approximately 30 mm is achieved.
A two axis gimbal less MEMS mirror allows raster, circular and
resonant scan patterns, which are not practical with a classical 2 axis
galvo scanner, because of heavy beam movement at the pupil of the
eye, caused by off-pivot point scanning.
Eyes of healthy subjects were imaged and the mean total retinal blood
flow as well as the velocity profiles inside all major retinal vessels
emerging from the ONH were extracted.
Results: Figure 1 shows an example of a circular scan around
the ONH of a healthy human subject. The velocity profiles of all
major vessels are visible. Furthermore 3 beam D-OCT allows the
reconstruction of the vessel geometry, showing excellent agreement
between the actual and calculated vessel orientation as well as the
flow direction. The total venous mean flow was 54.7 ml/min, while
the total arterial flow was 47.8 ml/min.
Conclusions: The improved three beam D-OCT technique allows
the direct measurement of total retinal blood flow as well as velocity
vector determination with various scan patterns independent from
any a-priory knowledge on the vessel geometry. With further
development the technique may aid the early diagnosis of eye
diseases like glaucoma.
Commercial Relationships: Makoto Araie, Topcon (P); Aiko
Iwase, Topcon (P); Ryo Kawasaki, None; Jun Suehiro, Topcon (E);
Akihiko Sekine, Topcon (E)
Conclusions: We quantified BF in the largest vessels and the
smallest capillaries in the mouse retina, which has been challenging
for other approaches that may lack the necessary spatial and
temporal resolution. The rapid rate of acquisition allowed precise
measurements of fast blood velocity fluctuations associated with
the mouse cardiac cycle, in arterioles, venules and capillaries.
This method can provide a complete analysis of hemodynamics
throughout retinal circulation, which is important to study ocular
diseases that disturb regional and global BF, such as diabetic
retinopathy and glaucoma.
Commercial Relationships: Richard Haindl, None; Wolfgang
Trasischker, None; Bernhard Baumann, None; Andreas Wartak,
None; Michael Pircher, None; Christoph K. Hitzenberger, None
Support: Austrian Science Fund Grant Number P26553-N20
Measurement of blood flow in the largest vessels and smallest
capillaries in the living mouse retina using an adaptive optics
scanning light ophthalmoscope
Aby Joseph2, 1, Andres Guevara-Torres2, 1, David R. Williams2, 1, Jesse
B. Schallek2. 1Institute of Optics, University of Rochester, Rochester,
NY; 2Center for Visual Science, University of Rochester, Rochester,
NY.
Purpose: Blood cell velocities in the mammalian retina range from
mm/s to cm/s, depending on vessel size and health. Here we automate
blood flow (BF) measurement from the full spectrum of vessels using
an adaptive optics scanning light ophthalmoscope (AOSLO).
Methods: Anesthetized C57BL/6J mice were imaged with an
AOSLO using near infrared light. Vessel lumen diameters were
measured in raster mode using sodium fluorescein contrast.
Additionally, single blood cells were imaged as they passed through
a one dimensional 31 kHz scanning beam (Zhong et al 2008). Images
of cell position as a function of time were captured; cells manifest as
bright streaks. Streak slope was automatically determined using the
Radon transform, which can measure cell velocities up to 1 m/s. BF
in each vessel was calculated assuming laminar flow and cylindrically
shaped vessels.
Results: Measured inner lumen diameters ranged from 4.5 mm in the
capillaries to 53 mm in the largest vessels near the disc. Measured
erythrocyte velocities ranged from stopped flow in capillaries to 35
mm/s in large vessels. Temporal variations in velocity, measured in
both arterioles and venules, corresponded to heart rates of 202–280
bpm, that matched well with simultaneous measurements with a pulse
oximeter.
The velocity signal was robust in vessels that intersected the scanning
beam at less than 40° (45% of vessels), the contrast between
erythrocyte streaks being insufficient for remaining vessels. BF in the
remaining vessels was estimated by scaling the measured BF in other
vessels by the fourth power of the ratio of the widths of each vessel
(Poiseuille’s Law). Averaging across time, total retinal BF values
ranging from 3.76–4.79 mL/min in arterioles and 3.85–4.78 mL/min in
venules were calculated.
Purpose: Geometric changes of the retinal vasculature may precede
or form part of the pathophysiology of eye diseases such as diabetic
retinopathy or retinal vascular occlusion. Quantitative vasculature
parameters are used clinically and in research for risk assessment,
disease monitoring and as markers of therapeutic efficacy. This
study describes a new method for rapid automated measurement of
vascular geometric parameters within a selectable region. Baseline
measurements for the adult C57BL/6J mouse retina are defined.
Methods: Retinal flat-mounts were prepared from adult C57BL/6
mice, stained with fluorophore-conjugated isolectin B4, and imaged
by confocal microscopy. Animal care guidelines of the ARVO
Statement for the Use of Animals in Ophthalmic and Vision Research
were followed and all procedures were approved by the institutional
animal care and use committee (St Vincent’s AEC protocol
SABC001). 8 retinal flat-mounts 4900×5800 pixels were obtained
and corrected for uneven illumination. Images were binarized using
an SVM classifier. Vessel bifurcations and crossover points were
automatically detected using a novel backward morphological
shrinking operation as shown in Figure 1. Flat-mount images were
partitioned into 5 non-overlapping regions 1200×1200 pixels centred
on the optic disc (OD) and each retinal quadrant (RQ). Vascular
parameters including the branching angles (value and count), vessel
to background (V/B) ratio and fractal dimension (FD), were obtained
for each region.
Results: FDs for OD (mean±SD=1.522 ± 0.012) and RQ (1.526 ±
0.021) regions were similar. Similarity was also found in V/B ratio
(OD=18.86 ± 1.54 %, RQ= 18.80 ± 2.83%). However, OD had fewer
branch points (bp) (255± 39.55) and acute angle(aa) (88.64± 24.08
degrees) compared to RQ (bp=355.5±97.71, aa =92.16 ± 21.09).
Conclusions: This novel technique provides a fast, automated and
reliable method for quantification of vascular parameters including
estimates of the mean and variance of branching angles which is not
feasible manually or by other software modules such as VESGEN
(VESsel GENeration Analysis. Also unlike VESGEN, our method
does not require a binary vascular image as the input and works with
any grayscale image format. The reliable quantification of retinal
vascular parameters is likely to be of value to studies of retinal
vascular disease, including angiogenesis, and putative therapies for
these disorders.
Commercial Relationships: Aby Joseph, None; Andres GuevaraTorres, Canon, Inc. (F), University of Rochester (P); David R.
Williams, Canon, Inc. (F), Canon, Inc. (R), Polgenix, Inc. (F),
University of Rochester (P); Jesse B. Schallek, University of
Rochester (P)
Support: Research reported in this publication was supported by
the National Eye Institute of the National Institutes of Health under
Award No. P30 EY001319 and F32 EY023496. The content is solely
the responsibility of the authors and does not necessarily represent
the official views of the National Institute of Health.
Automated measurement of vascular parameters in mouse retinal
flat-mounts
Behzad Aliahmad1, Dinesh K. Kumar1, Marc Sarossy2, Elsa Chan3,
Peter V. Wijngaarden3. 1RMIT University, Melbourne, VIC, Australia;
2
Royal Victorian Eye and Ear Hospital, Melbourne, VIC, Australia;
3
Centre for Eye Research Australia & Department of Ophthalmology,
The University of Melbourne, Melbourne, VIC, Australia.
vascular layers, including Choriocapillaris and often to highlight the
connections between these layers.
Conclusions: Our study shows the capability of the Spectralis OCT2ANGIOGRAPHY system to analyze in detail the whole retinal and
choroidal vascular tissue in order to be a useful reference in case of
pathological conditions.
Figure 1
Commercial Relationships: Behzad Aliahmad, None; Dinesh K.
Kumar, None; Marc Sarossy, None; Elsa Chan, None; Peter V.
Wijngaarden, None
Conventional Multimodal Imaging Versus “En-Face” OCT2
Angiography: deep inside the chorioretinal vascular tissue
Marco Lupidi1, 2, Gabriel J. Coscas2, Florence Coscas2. 1S.Maria
Della Misericordia Hospital, Eye Clinic, Perugia, Italy; 2Centre
Ophtalmologique De L’Odeon, Paris, France.
Purpose: To describe and compare the conventional multimodal
imaging findings to the images of all the retinal and choroidal
vascular layers obtained with the Spectralis “En-Face” OCT2ANGIOGRAPHY in healthy subjects without dye injection.
Methods: Prospective case series of 10 eyes of 10 consecutive
asymptomatic patients (7 females, mean age 34 ± 15) were evaluated
by the use of a conventional multimodal protocol including
Fluorescein Angiography (FA) and Enhanced Depth Imaging
(EDI) B-Scan Spectral Domain Optical Coherence Tomography
(OCT) and by a new multimodal approach based on the “En-Face”
visualization of the macular area obtained with a Spectralis OCT2ANGIOGRAPHY (Heidelberg Engineering, Heidelberg, Germany).
A macro-structural and micro-structural analysis (on a 15 x 10°
area) was performed to achieve the difference in detecting capability
between the two diagnostic protocols. In order to obtain the best
En-Face OCT2-ANGIOGRAPHY images a 25 mm thickness C-scan
section was shaped on the ILM profile till the end depth of the foveal
depression and then modeled to the BM shape from this level to the
Haller’s layer.
Results: The visualization of the superficial retinal vascular layers
was appreciable with both protocols, but the En-Face OCT2ANGIOGRAPHY offered the possibility to study in detail the
deep capillary plexus (in its two components) and the connections
with the superficial one. Moreover the En-Face and the OCT2ANGIOGRAPHY allowed us to distinguish different choroidal
Commercial Relationships: Marco Lupidi, None; Gabriel J.
Coscas, None; Florence Coscas, None
Characterizing the Effect of Anti-VEGF Injection on Choroidal
Neovascularization Using Optical Coherence Tomography
Angiography
Nora W. Muakkassa, Adam T. Chin, Talisa de Carlo, Caroline
R. Baumal, Andre Witkin, Jay S. Duker, Nadia K. Waheed.
Ophthalmology, New England Eye Center/Tufts Medical Center,
Boston, MA.
Purpose: To measure the change in size of choroidal
neovascularization (CNV) in response to anti-vascular endothelial
growth factor (VEGF) treatment measured using a spectral domain
optical coherence tomography angiography (SD-OCTA) system.
Methods: Eight eyes from eight patients diagnosed with CNV
who underwent OCTA pre and post-treatment with anti-VEGF
agents were enrolled. One eye was treatment-naïve. Each eye was
imaged prior to treatment and scanned at multiple follow up visits
using the prototype AngioVue OCTA system applying a splitspectrum amplitude decorrelation angiography (SSADA) algorithm
on a SD-OCT device (Optovue Inc, Fremont, CA). The system
operated at 70,000 A-scans per second to acquire OCTA volumes
of 304 x 304 A-scans. Orthogonal registration and merging of
two consecutive image sets were used to obtain OCT angiograms.
OCTA volumes were segmented into 30-50 micrometer-thick enface segments demonstrating the dimensions of the CNV. Trained
readers at the Boston Image Reading Center measured the greatest
linear dimension and the area of the CNV in the pre-treatment and
post-treatment images. The primary outcomes of the study were
greatest linear dimension and area of the CNV. Secondary outcomes
included the appearance of CNV (well-circumscribed versus poorly
circumscribed), the presence of subretinal and/or intraretinal fluid on
OCT B-scan, and visual acuity.
Results: Eight eyes of eight patients with CNV were scanned with
the OCTA system. Ages ranged from 31 – 85 and included four
males and four females. Etiology of CNV included age-related
macular degeneration, multifocal choroiditis and central serous
chorioretinopathy. 50 % of eyes had significant reduction in the
length and area of the CNV after treatment as well as resolution of
subretinal fluid. 25 % had no change in the size of the CNV and no
change in amount of subretinal fluid present.
Conclusions: CNV size measured using OCTA may decrease after
successful anti-VEGF treatment. No change in the size of the CNV
may suggest resistance to anti-VEGF agents. Optical coherence
tomography angiography may be a useful tool for monitoring and
quantifying the response of choroidal neovascularization to treatment.
Commercial Relationships: Nora W. Muakkassa, None; Adam T.
Chin, None; Talisa de Carlo, None; Caroline R. Baumal, None;
Andre Witkin, None; Jay S. Duker, Carl Zeiss Meditec Inc. (C),
Carl Zeiss Meditec Inc. (F), EyeNetra (I), Hemera Biosciences
Inc. (I), Ophthotech Corp (I), Optovue (C), Optovue (F); Nadia K.
Waheed, None
and Massachusetts Lions Club.
Methodology for visualization of reduced choriocapillaris density
using optical coherence tomography angiography
Simon S. Gao, Yali Jia, Nieraj Jain, Mark E. Pennesi, David Huang.
Ophthalmology, Oregon Health & Science University, Portland, OR.
Purpose: To develop a method of visualizing reduced
choriocapillaris density using optical coherence tomography (OCT)
angiography.
Methods: Macular scans of a participant with a clinical diagnosis
of choroideremia were taken using a spectral OCT system (RTVueXR) and compared with normal participants. The split-spectrum
amplitude-decorrelation angiography algorithm was used to detect
flow. The inner retinal layer from the internal limiting membrane
to 90 mm below showed the retinal circulation. The choriocapillaris
layer was defined to be 10 mm below Bruch’s membrane to 20 mm
below. Flow projections from retinal vessels were treated as null in
the choriocapillaris layer. The choriocapillaris angiogram was then
split into 8x8 superpixels, and the vessel density was calculated for
each superpixel. This vessel density map was then smoothed using
linear interpolation.
Results: In the choroideremia case, en face OCT angiogram of the
retinal circulation appeared normal (Fig. 1A). The en face structural
OCT of the retinal pigment epithelium (RPE) showed hyporeflective
regions indicating RPE loss (Fig. 1B). The choriocapillaris angiogram
and vessel density map showed distinct areas of reduced flow (Fig.
1C-E). Overlay (Fig. 1F) showed patches of reduced choriocapillaris
vessel density within regions of RPE loss.
Conclusions: We have developed a method to visualize reduced
choriocapillaris density. This method may be useful in the assessment
of diseases such as choroideremia and age-related macular
degeneration.
Figure 1. (A) En face OCT angiogram of the retinal circulation in
a patient with choroideremia. (B) En face OCT structural image
of the retinal pigment epithelium. (C) OCT angiogram of the
choriocapillaris. (D) Vessel density map of the choriocapillaris. (E)
Smoothed vessel density map. Scale bar shows vessel density from
0% (blue) to 100% (red). (F) Areas where the vessel density was less
than 50% (pink-purple) overlaid on (B).
Commercial Relationships: Simon S. Gao, None; Yali Jia,
Optovue, Inc. (F), Optovue, Inc. (P); Nieraj Jain, None; Mark E.
Pennesi, Sucampo Pharmaceuticals (C); David Huang, Carl Zeiss
Meditec, Inc. (P), Optovue, Inc. (F), Optovue, Inc. (I), Optovue, Inc.
(P)
Support: NIH grants R01 EY023285, R01 EY024544, DP3
DK104397, K08 EY021186 and T32 EY23211; CTSA grant
UL1TR000128; FBB grants CD-NMT-0914-0659-OHSU and
CDA CF-CL-0614-0647-OHSU; Alcon Research Institute Young
Investigator Grant; Career Development Award from RPB and an
unrestricted grant from RPB
Longitudinal monitoring of choroidal neovascularization by OCT
angiography in mice
wenzhong Liu1, Ji Yi1, Ronil S. Shah2, Brian Soetikno3, Amani A.
Fawzi3, Hao F. Zhang1, 2. 1Biomedical Engineering, Northwestern
University, Evanston, IL; 2Department of Ophthalmology,
Northwestern University, Evanston, IL; 3Feinberg School of
Medicine, Northwestern University, Evanston, IL.
Purpose: Early detection of choroidal neovascularization (CNV)
remains challenging in clinical practice by existing optical coherence
tomography (OCT) and fluorescein angiography. Recently developed
OCT angiography enhanced the blood flow contrast and suppressed
the other static tissue signals, which allows us to visualize CNV
directly. We hypothesize that OCT angiography can effectively detect
CNV.
Methods: Adult pigmented mice were anesthetized by ketamine/
xylene cocktail and subjected to laser treatment following established
protocols of laser induced CNV around optic disc. Two OCT systems
working in near infrared (NIR-OCT) and visible light range were
used to provide complementary images for CNV characterization. A
peripapillary area of 2x2mm2 centering optic nerve disk was imaged
with both OCT systems. The OCT angiography scanning protocol
scanned the same B-scan location twice and took the difference to
enhance the blood flow contrast. The choroidal vasculature was
extracted from the three-dimensional OCT angiography. OCT
structural image and angiography were registered and compared
between NIR and visible light OCT images. Four different CNV
lesions were monitored longitudinally for 4 weeks after laser.
Results: NIR-OCT provides better penetration depth into choroidal
layer (sample image can be found in Figure 1) and the visible light
OCT provides superb axial resolution down to 1 micrometer. We
quantified the CNV area/volume from 3D angiography. We observed
the choroidal vascular remodeling on both NIR and visible light
OCT, which started around 4 days and peeked one week after laser
induction. The CNV lesion gradually regressed afterwards.
Conclusions: Compared to conventional OCT images, OCT
angiography provides enhanced contrast from vasculature and can be
used to detect and monitor the progression of CNV.
Figure 1. NIR OCT angiography of a CNV mouse. (a) Sample OCT
angiography B-scan. The dash box indicates the CNV location. (b)
OCT angiography of major retinal vessels from the layer indicated by
black arrow in (a). (c) OCT angiography of retinal capillaries from
the layer indicated by purple arrow in (a). (d) OCT angiography of
choroidal vessels from the layer indicated by red arrow in (a). CNV
position is circled by the dash ring. Bar: 100 mm for (a) and 200 mm
for (b).
Commercial Relationships: wenzhong Liu, None; Ji Yi, None;
Ronil S. Shah, None; Brian Soetikno, None; Amani A. Fawzi,
None; Hao F. Zhang, None
Support: HHMI international student fellowship, NIH grants
1R01EY019951, NIH grants 1R24EY022883
Detection of occult choroidal neovascularization in age-related
macular degeneration with optical coherence tomography
angiography
Steven T. Bailey1, Yali Jia1, Christina J. Flaxel1, Thomas S. Hwang1,
Martin F. Kraus2, David Huang1. 1Ophthalmology, Casey Eye
Institute, OHSU, Portland, OR; 2Advanced Optical Technologies,
Pattern Recognition Lab, Erlangen-Nuremberg, Germany.
Purpose: To visualize occult choroidal neovascularization (CNV) in
age-related macular degeneration using OCT angiography.
Methods: 15 eyes from 13 study participants with active occult CNV,
as defined by fluorescein angiography (FA), were scanned with either
a prototype (100 kHz A-scan repetition rate) swept-source OCT
acquiring 3x3 mm scans or a commercially available high-speed
spectral OCT (70 kHz, RTVue XR, Optovue, Inc.) acquiring both
3X3 and 6X6 mm scans. Flow was detected with the split-spectrum
amplitude decorrelation angiography (SSADA) algorithm and motion
artifact was removed by 3D orthogonal registration and merging
of 2 scans. The volumetric angiogram was segmented into inner
retinal, outer retinal and choroidal layers to construct separate en face
angiograms. CNV was identified as vessels above Bruch’s Membrane
in the outer retinal layer. CNV area was assessed by summing
pixels with detectable flow in the en face outer retinal angiogram.
Composite color-coded angiograms allowed for en face visualization
of multiple vascular beds as well as representation of both flow and
structure on cross sections.
Results: In 9 eyes with fibrovascular pigment epithelial detachment
(PED), large CNV vessels with smaller vascular fronds were seen
with OCT angiography; while FA revealed only large diffuse
areas of staining and leakage (Fig 1). In 6 eyes with stippled
hyperfluorescence from an undetermined source with FA, OCT
angiography demonstrated a clearly defined vascular network in the
outer retina (Fig. 2). CNV area could be assessed in all cases. Crosssectional OCT angiograms demonstrated type 1 CNV in all cases.
Conclusions: OCT angiography with SSADA improved visualization
of occult CNV compared to FA. All cases were type 1 CNV. CNV
were ill-defined with FA due to fluorescein leakage and from
blockage from retinal pigment epithelial cells. CNV areas were
quantified with OCT angiography, which may be useful barometer in
assessing response to treatment.
Fig 1. A, Late phase FA. B, 6X6 mm OCT angiogram: retinal vessels
(purple) and CNV (yellow). C. Horizontal cross-sectional OCT
angiogram (retinal vessels – purple; CNV – yellow; choroidal vessels
– red) of type 1 CNV
Fig 2. A, Late phase FA. B, 3X3 mm OCT angiogram: retinal vessels
(purple) and CNV (yellow). C, Horizontal cross-sectional OCT
angiogram of type 1 CNV
Commercial Relationships: Steven T. Bailey, None; Yali Jia,
Optovue INC (P); Christina J. Flaxel, None; Thomas S. Hwang,
None; Martin F. Kraus, Optovue Inc. (P); David Huang, Carl Zeiss
Meditec, Inc. (P), Optovue, Inc. (C), Optovue, Inc. (F), Optovue, Inc.
(I), Optovue, Inc. (P), Optovue, Inc. (R)
Support: R01 EY024544, RPB, CTSA grant (UL1TR000128)
Sensitivity and Specificity of Ultra-High Speed Optical
Coherence Tomography Angiography in Detection of Choroidal
Neovascularization in Neovascular Age-Related Macular
Degeneration
Anthony Joseph1, Talisa E. de Carlo1, Mehreen Adhi1, Eric Moult2,
Nadia Waheed1, Caroline Baumal1, Woo Jhon Choi2, James G.
Fujimoto2, Jay S. Duker1, Andre Witkin1. 1New England Eye Center
and Tufts Medical Center, Boston, MA; 2Department of Electrical
Engineering and Computer Science, Massachusetts Institute of
Technology, Boston, MA.
Purpose: To estimate the sensitivity and specificity for the detection
of choroidal neovascularization (CNV) using ultrahigh speed swept
source optical coherence tomography angiography (SS-OCTA) in
patients with neovascular age-related macular degeneration (AMD).
Methods: In this retrospective observational cross-sectional
study, patients with a clinical diagnosis of neovascular AMD were
evaluated at New England Eye Center between January 2014 and
November 2014. 25 patients who received ultrahigh speed SSOCTA (Massachusetts Institute of Technology, Cambridge MA)
and 50-degree fluorescein angiography (FA) on the same day were
evaluated. The SS-OCTA software co-registers the SS-OCTA enface angiograms with SS-OCTA correlating b-scans allowing for
visualization of structure and blood flow in tandem. FA, SS-OCTA
angiograms, and SS-OCTA correlating b-scans were independently
reviewed by two masked Boston Image Reading Center (BIRC)
trained readers to identify CNV. Sensitivity and specificity of CNV
detection using SS-OCTA were calculated using FA as the ground
truth in three different ways: (1) SS-OCTA angiograms alone (2)
SS-OCTA correlating b-scans alone (3) SS-OCTA angiograms and
correlating b-scans.
Results: Sensitivity of CNV detection was moderate when using
either SS-OCTA angiograms or SS-OCTA correlating b-scans alone.
However, sensitivity improved when using the SS-OCTA images in
tandem. Specificity was high in all three groups.
Conclusions: SS-OCTA is able to non-invasively visualize CNV
which may provide a method for identifying and guiding treatment of
CNV. The sensitivity and specificity of CNV detection on ultra-high
SS-OCTA compared with FA appears high. Future studies with larger
sample sizes are needed to better elaborate on the sensitivity and
specificity of CNV detection and illustrate clinical utility.
Commercial Relationships: Anthony Joseph, None; Talisa E.
de Carlo, None; Mehreen Adhi, None; Eric Moult, None; Nadia
Waheed, None; Caroline Baumal, None; Woo Jhon Choi, None;
James G. Fujimoto, Carl Zeiss Meditech Inc. (P), Optovue Inc.
(I), Optovue Inc. (P); Jay S. Duker, Carl Zeiss Meditech Inc. (C),
EyeNetra (I), Hemera Biosciences Inc. (I), Ophthotech Corp. (I),
Optovue Inc. (C); Andre Witkin, None
Support: This work was supported by the National Institute
of Health (NIH R01-EY011289-27, R44-EY022864-01, R44EY022864-02, R01-CA075289-16), Air Force Office of Scientific
Research (AFOSR FA9550-10-1-0551 and FA9550-12-1-0499), a
Samsung Scholarship, and by a Natural Sciences and Engineering
Research Council of Canada Scholarship.
Histopathological correlation of optical coherence tomography
angiography in laser-induced choroidal neovascularization
Ronil S. Shah2, Brian Soetikno2, 1, wenzhong Liu1, Ji Yi1, Hao F.
Zhang1, Amani A. Fawzi2. 1Biomedical Engineering, Northwestern
Univeristy, Evanston, IL; 2Ophthalmology, Feinberg School of
Medicine - Northwestern Univeristy, Chicago, IL.
Purpose: In-vivo assessment of choroidal neovascularization (CNV)
is currently done via optical coherence tomography (OCT) and
fluorescein/ICG angiography, and recently, the novel technology of
OCT angiography (OCT-ang) - a system capable of measuring retinal
blood flow. The purpose of this study is to explore whether OCTang accurately correlates with histological neovascularization in an
experimental model of murine laser induced CNV.
Methods: Twenty adult pigmented mice of mixed background were
anesthetized and subjected to laser induced CNV following preestablished protocols. Four laser lesions were applied around the
optic nerve to induce rupture Bruch’s membrane; laser applications
that did not result in a bubble formation indicating rupture of Bruch’s
membrane were excluded from analysis. Animals were imaged
with OCT-ang at varying intervals between 1 to 28 days post laser
injury in order to visualize the development & regression of CNV.
Histological samples were obtained at the corresponding intervals –
eyes were paraformaldehyde-fixed and stored in phosphate buffered
saline at 4oC. The choroid was dissected, stained with isolectin B4
and flat mounted to visualize the CNV. These images were compared
to OCT-ang images obtained the same day.
Results: Comparing the images obtained from OCT-ang to the
immunostained flat mount shows consistency between the two
modalities. The vessels visualized by the OCT-ang are representative
of the vessels visualized via immunostaining of the choroidal flat
mount (see figure). We observed that the OCT-ang images have
better contrast between background and neovascular images. Further
analysis of the various time points is underway to identify the earliest
detectable lesion on OCT-ang.
Conclusions: OCT angiography of CNV correlates well with the true
choroidal morphology as viewed in choroidal vascular flat mount.
Since the murine laser induced CNV model is a widely accepted
model for human neovascular AMD, this technology holds promise
for in vivo detection & management for neovascular AMD.
A) En face choroidal OCT-ang image of CNV lesion; B) Confocal
microscopy of choroid flat mount of the same lesion obtained
the same day - immunostained with isolectin B4 (stains vascular
endothelium). Arrows highlight same areas on both images; slight
disparity because OCT-ang image is a 20-micron thick slab compared
to a 2-micron thick confocal slice.
Commercial Relationships: Ronil S. Shah, None; Brian Soetikno,
None; wenzhong Liu, None; Ji Yi, None; Hao F. Zhang, None;
Amani A. Fawzi, None
Support: Research to Preventing Blindness, NY (Department of
Ophthalmology, Northwestern University); Macula Society Research
Grant (AAF)
Fate of non-perfused vessels in ischemic retina
Marcus Fruttiger1, Michael Powner1, Ryan Jones1, Weijen Tan1,
Meidong Zhu2, Andrew A. Chang3, Dawn A. Sim5, Pearse A. Keane4,
Adnan Tufail5, Catherine A. Egan5. 1UCL Institute of Ophthalmology,
London, United Kingdom; 2Save Sight Institute, Sydney, NSW,
Australia; 3Sydney Retina Clinic and Day Surgery, Sydney, NSW,
Australia; 4NIHR Biomedical Research Centre for Ophthalmolgy,
Moorfields Eye Hospital, London, United Kingdom; 5Moorfields Eye
Hospital, London, United Kingdom.
Purpose: Ischemic retinopathy is associated with several vision
threating complications, such as neural atrophy, vascular leakage
and neovascularisation. Traditionally ischemia has been assessed by
fluorescein angiography, visualising perfused vessels. Although this
method does not provide any information about non-perfused vessels,
it is often assumed that vessels in ischemic areas regress. Here we
aim to learn more about the longterm fate of non-perfused vessels in
the retinal vasculature.
Methods: Optical coherence tomography (Avanti Angiovue SDOCT,
Optovue, Inc. Fremont, CA, USA) was used to visualise perfusion
as well as structural properties of the retinal vasculature in patients
suffering from retinal vascular occlusions. In addition, post mortem
tissue from a patient with long standing (6 years) central retinal vein
occlusion (CRVO) was investigated, using immunohistochemistry
on whole mount retina and paraffin sections to visualise blood vessel
components and retinal cells.
Results: Comparing OCT angiography (based on speckle variance)
with en-face OCT images from selected retinal layers revealed that
in ischemic areas of the retina non-perfused, larger vessels could
be detected as hyper reflective structures. Furthermore, analysis of
the CRVO postmortem tissue revealed perfect preservation of the
basement membrane from all retinal vessels, including capillaries.
However, these non-perfused “vessels sleeves” did not contain
endothelial cells or pericytes.
Conclusions: Our data suggests longterm preservation of vascular
basement membrane in ischemic retina. This has implications for
therapeutic approaches aiming to alleviate retinal ischemia via cell
therapy.
Commercial Relationships: Marcus Fruttiger, Astra Zeneca (F),
Novartis (F); Michael Powner, None; Ryan Jones, None; Weijen
Tan, None; Meidong Zhu, None; Andrew A. Chang, None; Dawn
A. Sim, None; Pearse A. Keane, None; Adnan Tufail, Alergan (C),
Bayer (C), GSK (C), Novartis (C), Pfizer (C), Thrombogenics (C);
Catherine A. Egan, None
Support: Fight for Sight, Diabetes UK, NIHR Biomedical
Research Centre at Moorfields Eye Hospital and UCL Institute
of Ophthalmology, Moorfields Trustees, Lowy Medical Research
Institute
Imaging of choroidal neovascularization by Angio-OCT: a
comparison with indocyanine green angiography (ICGA)
Vittoria Ravera, Marco Pellegrini, Marta Oldani, Matteo G. Cereda,
Alessandra Acquistapace, Giovanni Staurenghi. Clinical Science
“Luigi Sacco”, University of Milan, Milan, Italy.
Purpose: To evaluate the visibility of choroidal neovascularizations
(CNVs) in patients affected by age related macular degeneration
(ARMD) or inflammatory affections by Angio-OCT imaging
and to compare findings with fluorescein and indocyanine green
angiography (FA and ICGA)
Methods: Retrospective study. All the patients underwent a complete
ophthalmological examination including blue autofluorescence
(B-FAF), fluorescein (FA) and indocyanine green (ICGA)
angiography, spectral domain optical coherence tomography (SDOCT) (HRA + OCT Spectralis, Heidelberg Engineering, Heidelberg,
Germany) and angio-OCT using AngIoVue technologies (Optovue
Inc.)
Results: 20 eyes of 20 patients with CNVs were included in the
study: 19 secondary to exudative age-related macular degeneration
and one inflammatory. FA was evaluated in order to classify CNVs
and to define their activity. 13 eyes displayed CNV type 1 (occult),
4 CNV type 2 (classic), 2 presented a combination of the two
(minimally or predominantly classic), and 1 was a CNV type 3
(retinal angiomatous proliferation). On ICGA, CNV complex was
visible in 99% of cases, whereas the precise characterization of the
vessels belonging to the lesion was possible only in 30% of cases.
On Angio-OCT the visibility of the CNV complex was typically
better compared with ICGA (100% of cases), particularly allowing
a better identification of the neovascular network. On SD-OCT we
determined the presence of subretinal fluid or blood and pigmented
epithelium detachment (PED) to determine if they reduced the
visibility of the net on SSADA. Imaging was not limited by either
subretinal fluid or blood while PED produced mild masking
Conclusions: Angio-OCT represents a novel diagnostic technique
capable to perform non-invasive high resolution imaging of the retina
and choroid. Further studies are required to study its reproducibility
and applications in the clinical scenario
Commercial Relationships: Vittoria Ravera, None; Marco
Pellegrini, Bayer (S); Marta Oldani, None; Matteo G. Cereda,
None; Alessandra Acquistapace, None; Giovanni Staurenghi,
Alcon Laboratories, Inc. (C), Allergan, Inc. (C), Bayer (C),
Boehringer (C), Genentech (C), GlaxoSmithKline (C), Heidelberg
Engineering (C), Novartis Pharmaceuticals Corporation (C), Novartis
Pharmaceuticals Corporation (S), Ocular instruments, Inc. (P), Optos,
Inc. (C), Optovue (S), Roche (C), Zeiss (C), Zeiss (S)
OCT Angiography (OCTA) of Macular Neovascularization
(MNV)
Emeline R. Ramenaden1, John E. Legarreta1, Andrew D. Legarreta1,
Douglas Matsunaga2, Amir H. Kashani2, Giovanni Gregori1, Qinqin
Zhang3, Ruikang K. Wang3, Carmen A. Puliafito2, Philip J. Rosenfeld1.
1
Ophthalmology, Bascom Palmer Eye Institute, University of Miami
Miller School of Medicine, Miami, FL; 2USC Eye Center, Keck
School of Medicine, University of Southern California, Los Angeles,
CA; 3Bioengineering, University of Washington, Seattle, WA.
Purpose: To evaluate the microvasculature of the central macula in
eyes with macular neovascularization (MNV) using swept-source
(SS) and spectral-domain (SD) optical coherence tomography (OCT)
angiography.
Methods: Subjects were enrolled in a prospective, observational
study and evaluated using a high-speed 1050 nm SS-OCT prototype
system (100,000 kHz) and a 840 nm SD-OCT prototype system
(68,000 kHz). SS-OCT angiography was performed using a 3X3mm
raster scan pattern centered on the fovea. In the transverse scanning
direction, a single B-scan was comprised of 300 A-scans. Four
consecutive B-scans were performed at each fixed position before
proceeding to the next transverse position on the retina. A total of
300 B-scan positions located 10 μm apart over a 3 mm distance were
sampled. SD-OCT angiography was performed using a 3X3mm and
6X6mm raster scan pattern. In the 3X3 raster scan, four consecutive
B-scans, each comprised of 245 A-scans, was performed in the
transverse scanning direction. In the 6X6 raster scan pattern, the
transverse scanning direction was comprised of two consecutive
B-scans, each of 350 A-scans. A total of 245 B-scan positions
were located 12.4 μm apart over the 3 mm distance, and a total of
350 B-scan positions were located 17.1 μm apart over the 6 mm
distance. Algorithms segmented the retina into three layers; an inner
retinal layer, a middle retinal layer, and an outer retinal layer. The
choriocapillaris and choroidal vasculature were further segmented.
The vascular distribution in each layer was depicted as an en face
image. En face OCT angiographic images were compared to early
and late phase fluorescein angiography (FA) images.
Results: Imaging was performed using the SS-OCT and SD-OCT
instruments. OCT angiography of MNV showed well-defined
microvascular networks. Furthermore, imaging before and after
treatment with aflibercept demonstrated a decrease in size of the
microvascular network after therapy. The decreased size of the
neovascular network exposed marked flow impairment in the
underlying choriocapillaris. Images obtained with both the SS-OCT
and SD-OCT instruments are being compared in this ongoing study.
Conclusions: OCTA, which can be performed with both SS-OCT
and SD-OCT instruments, provides rapid, non-invasive, highresolution, depth-resolved, images comparable to or even better
than conventional fluorescein angiography in eyes with macular
neovascularization.
Commercial Relationships: Emeline R. Ramenaden, Carl Zeiss
Meditec, Inc (F); John E. Legarreta, Carl Zeiss Meditec, Inc
(F); Andrew D. Legarreta, Carl Zeiss Meditec, Inc (F); Douglas
Matsunaga, Carl Zeiss Meditec, Inc (F); Amir H. Kashani, Carl
Zeiss Meditec, Inc (F); Giovanni Gregori, Carl Zeiss Meditec, Inc
(F), Carl Zeiss Meditec, Inc (P); Qinqin Zhang, Carl Zeiss Meditec,
Inc (F); Ruikang K. Wang, Carl Zeiss Meditec, Inc (F); Carmen
A. Puliafito, Carl Zeiss Meditec, Inc (F); Philip J. Rosenfeld, Carl
Zeiss Meditec, Inc (F)
Support: Research Support from Carl Zeiss Meditec, Inc
intensity based algorithm and visualized en face for comparison with
fluorescein angiograms (FA).
Results: OCTA in subjects with non-proliferative diabetic
retinopathy (DR) or proliferative DR showed areas of non-perfusion,
irregular capillaries, and microaneurysms that were qualitatively
similar to findings on FA. However, capillary non-perfusion could
be localized to either the superficial or middle retina in most cases
using OCTA whereas this distinction could not be made with FA.
Microaneurysms were observed in OCTA but their number, size and
shape did not correspond to FA findings in some cases.
Conclusions: OCTA generates high-resolution images that are
qualitatively similar to retinal vasculature imaged with conventional
fluorescein angiography. While OCTA is completely non-invasive
and may be performed with greater ease and frequency than
conventional fluorescein angiography, some clinical findings, such
as vascular leakage, cannot yet be assessed with OCTA. OCTA may
serve as an alternative method of assessing retinal vascular changes
when conventional fluorescein angiography cannot be performed.
OCT Angiography (OCTA) of Diabetic Retinopathy
Douglas Matsunaga1, Jack Yi1, Lisa C. Olmos1, John Legarreta3,
Andrew D. Legarreta3, Giovanni Gregori3, Utkarsh Sharma2,
Philip J. Rosenfeld3, Carmen A. Puliafito1, Amir H. Kashani1. 1USC
Eye Center, Keck School of Medicine of USC, Los Angeles, CA;
2
Research and Development, Carls Zeiss Meditec, Dublin, CA;
3
Bascom Palmer Eye Institute, Miami, FL.
Purpose: To non-invasively evaluate the retinal microvasculature in
diabetic human subjects with swept-source (SS) and spectral-domain
(SD) optical coherence tomography (OCT) angiography.
Methods: Ten subjects diagnosed with diabetes mellitus were
enrolled in a cross-sectional, observational study and all underwent
a complete ophthalmic examination. Subjects were evaluated using
a high-speed 1050 nm SS-OCT prototype system (100,000 kHz)
and a 840 nm SD-OCT prototype system (67,500 kHz). SD-OCT
angiography was performed using a 3X3mm and 6X6mm raster
scan pattern, which consisted in the transverse scanning direction
of, respectively, four consecutive B-scans each comprised of 245
A-scans in the 3X3 raster scan and a pair of consecutive B-scans
each comprised of 350 A-scans in the 6X6 raster scan pattern. A
total of 245 B-scans positions located 12.4 μm apart over the 3 mm
distance and 350 B-scans positions located 17.1 μm apart over the
6 mm distance were sampled. Retinal vasculature was assessed
in three retinal slabs consisting of the superficial retina, middle
retina and outer retina. The vasculature was reconstructed using an
Inner retinal OCTA and FA of the fovea in a subject with severe nonproliferative diabetic retinopathy.
Inner retinal OCTA and FA of the fovea in a subject with severe nonproliferative diabetic retinopathy.
Commercial Relationships: Douglas Matsunaga, Carl Zeiss
Meditec (F); Jack Yi, Carl Zeiss Meditec (F); Lisa C. Olmos, Carl
Zeiss Meditec (F); John Legarreta, Carl Zeiss Meditec (F); Andrew
D. Legarreta, Carl Zeiss Meditec (F); Giovanni Gregori, Carl Zeiss
Meditec (F); Utkarsh Sharma, Carl Zeiss Meditec (E); Philip J.
Rosenfeld, Carl Zeiss Meditec (F); Carmen A. Puliafito, Carls Zeiss
Meditec (F); Amir H. Kashani, Carls Zeiss Meditec (F)
Support: An Unrestricted grant from Research to Prevent Blindness,
New York, NY 10022
OCT-based microangiography of diabetic retinopathy
Qinqin Zhang1, Cecilia S. Lee2, Yanping Huang1, Kasra AttaranRezaei2, Jennifer R. Chao2, Richard Munsen2, James L. Kinyoun2,
Ruikang K. Wang1, 2. 1Department of Bioengineering, University of
Washington, Seattle, WA; 2Department of Ophthalmology, University
of Washington, Seattle, WA.
Purpose: To perform a feasibility study using OCT-based
microangiography (OMAG) to generate detailed retinal
microvascular maps in human subjects with diabetic retinopathy
(DR)
Methods: A 67 kHz Cirrus HD-OCT prototype system with motion
tracking (Carl Zeiss Meditec, Dublin, CA) was used to generate
retinal microvascular maps in 20 eyes of 10 subjects with DR by the
use of OMAG algorithms. Phase compensation and cross-correlation
methods were used to provide high quality capillary perfusion maps
of the retina from the 3D dataset. We also used a segmentation
algorithm to separate microvasculature within landmarked
physiological depth layers of the retina as well as within the choroid.
The layers include ganglion cell layer+ inner plexiform layer (Inner
retinal layer), inner nuclear layer + outer plexiform layer (middle
retinal layer) and outer nuclear layer (Outer retinal layer). For better
visualization, color coding was used for different layers: red —
inner retinal layer; green — middle retinal layer and blue — outer
retinal layer. Clinical fundus images and fluorescein angiograms (for
most subjects) were acquired on the same day as the OCT scan and
compared with resulting enface OMAG images
Results: Retinal OMAG images generated from the Cirrus-5000
system showed clearer microvascular maps compared to fundus
images and FA. In addition, OMAG provides depth-resolved
information, allowing visualization of the three landmarked
physiological layers. The microvascular features of typical DR
observed include microaneurysms, capillary dropout, and dilated or
tortuous vessels. The results also demonstrated the ability of OCTangiography to distinguish different forms of microaneurysms, such
as saccular, fusiform and focal bulges. Most of the microaneurysms
(green dots) were observed in the deeper retinal capillary plexus,
originating from locations between INL and OPL. Tortuous vessels
were also observed in the temporal-superior region, which is a
significant feature of DR. Fig.1 gives an example of an OMAG
depiction of irregular enlargement of the foveal avascular zone,
consistent with mild diabetic macular ischemia, compared with the
corresponding fundus image
Conclusions: The OCT-angiography prototype system demonstrated
the capability to generate detailed retinal OMAG microvascular maps
that can be valuable in aiding the diagnosis and perhaps treatment of
DR
Fig 1 OMAG provides detailed visualization of retinal vasculature in
DR patient
Commercial Relationships: Qinqin Zhang, Carl Zeiss Meditec
(F); Cecilia S. Lee, None; Yanping Huang, None; Kasra AttaranRezaei, None; Jennifer R. Chao, None; Richard Munsen, None;
James L. Kinyoun, None; Ruikang K. Wang, Carl Zeiss Meditec
(F), Carl Zeiss Meditec (P)
Prevent Blindness.
Diagnosis of nonproliferative diabetic retinopathy by
microaneurysm detection on swept source optical coherence
tomography (SS-OCT)
Theodore Leng, Ryan W. Nelson. Ophthalmology, Byers Eye Institute
at Stanford, Stanford University School of Medicine, Palo Alto, CA.
Purpose: To describe a novel method of identifying retinal vascular
microaneurysms (MAs) in nonproliferative diabetic retinopathy
(NPDR) using swept source optical coherence tomography (SS-OCT)
Methods: SS-OCT images were acquired in 17 eyes with NPDR
using a prototype SS-OCT device with a laser wavelength centered at
1060 nm and an acquisition speed of 100,000 A-scans/sec. 3 x 3 x 3
mm raster scans were obtained centered on the fovea (512 A-scans/Bscan, 512 B-scans/cube, 1500 pixels of depth). Sequential restricted
summed voxel projections, or “slabs,” were created with a thickness
of 4 μm through the cube and the images registered with intravenous
fluorescein angiography (FA) images obtained at the same visit. MAs
were identified on SS-OCT slabs and correlated to MAs identified on
FA images.
Results: MAs were identified in SS-OCT slabs in 15 of 17 eyes,
resulting in a NPDR diagnosis rate of 88%.
A mean of 20.9 slabs (SD 3.0) were analyzed in each eye. The mean
number of MAs identified on each FA was 11.7 (SD 11.9, range
1-38). The mean number of MAs identified via SS-OCT slabs was 8.1
per cube (SD 9.3, range 0-30); 62.7% (SD 31, range 0-100). The two
cases with no SS-OCT MA detection had only one MA identified on
FA. Ultimately, 68.84% of MAs were identified via SS-OCT slabs.
Conclusions: SS-OCT visualization of MAs could serve as a tool
for the diagnosis of NPDR. This technique should be explored in
larger studies. It may also be possible to apply this SS-OCT imaging
biomarker for population-based diabetic retinopathy screening
initiatives.
Fluorescein angiogram (FA) (left) and corresponding SS-OCT slab
image (right) of a right eye with nonproliferative diabetic retinopathy.
The right image corresponds to the area demarcated with the red box
on the FA image. Microaneurysms are noted with arrows on both
images.
Commercial Relationships: Theodore Leng, Carl Zeiss Meditec,
Inc. (C); Ryan W. Nelson, None
Support: The prototype SS-OCT device was provided by Carl Zeiss
Meditec, Inc. for this study. Institutional review board approval was
obtained as this device is not currently FDA certified, but 510K
clearance is pending.
Diabetic Retinopathy Features Detected with 6x6 mm OCT
Angiogram Using SSADA Algorithm
Thomas S. Hwang, Yali Jia, Simon S. Gao, Andreas K. Lauer,
Christina J. Flaxel, Steven Bailey, Phoebe Lin, David J. Wilson,
David Huang. Ophthalmology, Casey Eye Institute, Portland, OR.
Purpose: Diabetic retinopathy (DR) is a common retinal vascular
disease with classic fluorescein angiography (FA) features that
are useful for treatment and classification. OCT Angiography is a
dye-free alternative to FA. Using a commercially available 70kHz
OCT and the split-spectrum amplitude decorrelation angiography
(SSADA) algorithm, 6x6 mm images can be obtained in 3.5 seconds.
We evaluated the ability of this technique to detect the angiographic
features catalogued by the Early Treatment of Diabetic Retinopathy
Study (ETDRS).
Methods: Four patients with DR were imaged with conventional FA
and OCT angiography. The images were evaluated for the ability to
detect ETDRS features and other manifestations of vasculopathy.
Results: OCT angiography detected enlargement and distortion of
the foveal avascular zone, retinal capillary dropout, and pruning of
arteriolar branches. Areas of capillary loss obscured by fluorescein
leakage on FA were more clearly defined on OCT angiography. Some
areas of focal leakage on FA that were thought to be microaneurysms
were found to be small small tufts of neovascularization that
extended above the inner limiting membrane.
Conclusions: OCT angiography does not show leakage, but can
better delineate areas of capillary dropout and detect early retinal
neovascularization. This new noninvasive angiography technology
may be useful for routine surveillance of proliferative and ischemic
changes in diabetic retinopathy.
Figure 1 shows 6x6 mm fluorescein angiography (A) OCT
angiogram (B) of a patient with diabetic retinopathy with ETDRS
grid superposed showing FAZ enlargement inferotemporally and
temporally between the 300 (dotted) and 500 mm circles. Flow
signal detected between retinal pigment epithelium (RPE) and the
internal limiting membrane (ILM) is shown in magenta and signal
internal to the ILM is displayed in yellow. Magnified FA (C) shows
corresponding FAZ enlargement.
Figure 2: OCT angiogram (A) and FA (B) discloses areas of capillary
dropout in the temporal macula with pruning of the arterioles. In the
FA, diffuse leakage obscures an area of capillary drop out seen on
OCT angiography (red arrows). An arteriole with vessel wall staining
and narrowing (blue arrow) in the FA is shown to be a barely visible
ghost vessel on OCT angiography. Focal areas of leakage near the
fovea thought to be large microaneurysms on FA were shown to be
NV on OCT angiography (green arrows).
Commercial Relationships: Thomas S. Hwang, None; Yali Jia,
Optovue, Inc. (F), Optovue, Inc. (P); Simon S. Gao, None; Andreas
K. Lauer, Oxford BioMedica (C); Christina J. Flaxel, None; Steven
Bailey, None; Phoebe Lin, None; David J. Wilson, None; David
Huang, Carl Zeiss Meditec, Inc. (P), Optovue, Inc. (F), Optovue, Inc.
(I), Optovue, Inc. (P)
Support: DP3 DK104397, RPB, CTSA grant (UL1TR000128), T32
EY23211
Clinical evaluation of vascular lesions in diabetic retinopathy
using optical coherence tomography angiography
Akihiro Ishibazawa, Taiji Nagaoka, Atsushi Takahashi, Tsuneaki
Omae, Tomofumi Tani, Kenji Sogawa, Harumasa Yokota, Akitoshi
Yoshida. Ophthalmology, Asahikawa Medical University, Asahikawa,
Japan.
Purpose: To evaluate the ability of optical coherence tomography
angiography (OCTA) to visualize clinical fundus findings in patients
with diabetic retinopathy (DR).
Methods: Forty-seven eyes of 25 patients with DR were scanned
using a high-speed 840-nm-wavelength spectral-domain optical
coherence tomography (OCT) instrument (RTVue XR Avanti,
Optovue Inc., Fremont, CA). Blood flow was detected using the splitspectrum amplitude-decorrelation angiography (SSADA) algorithm,
and three-dimensional macular or optic disc angiography was
computed. Color fundus and fluorescein angiography (FA) images
also were obtained in all eyes and compared to the en-face SSADA
images for the ability to visualize microaneurysms (MAs), retinal
nonperfusion (RNP), and neovascularization (NV).
Results: Ninety-three percent of MAs detected by FA in 42 eyes
appeared as focally dilated saccular or fusiform capillaries in the
en-face SSADA images. The MAs were located in the superficial
vascular plexus (SVP) (30.3%) or the deep capillary plexus (DCP)
(69.7%). The RNP visualized by FA appeared as lesions with no
or sparse capillaries in the SSADA images. Measurement of the
RNP area near the macula in seven eyes showed that the RNP area
in the SVP (mean ± standard error of the mean [SEM], 3.67 ± 1.69
mm2) was larger than in the DCP (mean ± SEM, 3.02 ± 1.44 mm2).
The NV at the disc seen in four eyes on FA images and had marked
fluorescein leakage in the early phase. The vascular structure of the
NV was clearly visualized in the SSADA images. NV regression
and regeneration were quantified in an eye treated with antivascular
endothelial growth factor therapy.
Conclusions: OCTA can clearly visualize MAs and RNP and enables
evaluation of the retinal capillaries layer by layer. Quantitative
information on NV also can be obtained. OCTA may be clinically
useful to evaluate the microvascular status and therapeutic effect of
treatments for DR.
Commercial Relationships: Akihiro Ishibazawa, None; Taiji
Nagaoka, None; Atsushi Takahashi, None; Tsuneaki Omae, None;
Tomofumi Tani, None; Kenji Sogawa, None; Harumasa Yokota,
None; Akitoshi Yoshida, None
Support: Grant-in-Aid for Young Scientists (B) 25861608
Quantitative and qualitative evaluation of Diabetic Retinopathy
retinal vasculature with Cirrus-5000 Angiography prototype
Lin An1, Mary K. Durbin1, Scott Lee2, Patty Chung2, Michal Laron1,
Utkarsh Sharma3. 1Application and Clinical Department, Carl Zeiss
Meditec, Walnut Creek, CA; 2East Bay Retina Consultants, Inc.,
Oakland, CA; 3Advanced Development, Carl Zeiss Meditec, Dublin,
CA.
Purpose: To demonstrate that the new Cirrus-5000 Angio-prototype
is capable to achieve both quantitative and qualitative evaluations
of retinal vasculature features for human subjects with Diabetic
Retinopathy.
Methods: A Cirrus-5000 instrument was modified to allow OCT
angiography imaging. The system could provide two scanning
modes for imaging ocular vasculature, a 6 mm x 6 mm with 350 by
350 A-scans and a 3 mm x 3 mm scan with 245 by 245 A-scans to
achieve higher resolution. Both scanning modes could be finished ~
4 seconds. The microvasculature data obtained from OCT intensity
cube were segmented into three layers (superficial, deeper and outer
retinal layers) and color encoded into different colors (red, green
and blue respectively). The color composite image could be used to
indicate the relative depth positions of vasculature features.
A DR subject and a normal subject were recruited and imaged.
Fluorescein angiography imaging and OCT angiography imaging
were both performed on the DR subject, and the normal was imaged
only with OCTA. Visual comparison was performed between the FA
and OCTA images and features of interest were correlated. A vessel
density calculation method was applied on both DR and normal OCT
angiography images.
Results: OCT angiography revealed much clearer microvascular
details compared to the FA image. Many vasculature features were
well observed in the OCT angiography image. Microaneurysm,
capillary drop out and capillary tortuosity could be detected through
OCT angiography without using imaging dye. Unlike FA images, the
OCT angiography image provided depth resolved information.
The vessel density results of normal and DR subjects clearly
demonstrate that the vessel density of DR subject is lower than
normal subject, which is primarily due the capillary drop out of DR
subject.
Conclusions: The Cirrus-5000 Angio-prototype is capable of
achieving non-invasive detailed depth resolved microvasculature
map for DR retinal vasculature evaluation. The images delivered by
Cirrus-5000 Angio-prototype have good correlation with FA images,
quantitative analysis is also possible.
Fig 1. OCT angiography obtained from Cirrus-5000 angio prototype
not only has good correlation with FA image but also reveal better
details, including the micro-vasculature features and depth resolved
information.
Fig 2. Retinal Vessel density map of the normal (a) and DR subject
(b).
Commercial Relationships: Lin An, Carl Zeiss Meditec. Inc. (E);
Mary K. Durbin, Carl Zeiss Meditec. Inc. (E); Scott Lee, Carl Zeiss
Meditec. Inc. (C); Patty Chung, None; Michal Laron, Carl Zeiss
Meditec. Inc. (E); Utkarsh Sharma, Carl Zeiss Meditec. Inc. (E)
Prototype Ultra-High Speed Swept Source Optical Coherence
Tomography Angiography compared with Intravenous
Fluorescein Angiography in Diabetic Retinopathy
David A. Salz1, Talisa de Carlo1, 2, Mehreen Adhi1, 2, Eric Moult2, Woo
Jhon Choi2, Caroline R. Baumal1, Andre J. Witkin1, Jay S. Duker1,
James G. Fujimoto2, Nadia K. Waheed1. 1Ophthalmology, Tufts
Medical Center, Boston, MA; 2Department of Electrical Engineering
and Computer Science, and Research, MIT, Cambridge, MA.
Purpose: To compare the utility of a prototype ultrahigh speed swept
source optical coherence tomography (OCT) angiography with
intravenous fluorescein angiography (IVFA) in patients with diabetic
retinopathy (DR).
Methods: This is a prospective observational cross-sectional study.
A prototype ultra high speed swept source OCT angiography was
performed using 400,000 A scans per second to scan 3mm × 3mm
and 6mm × 6mm areas centered at the fovea in 11 normal eyes
and 29 eyes in patients with diabetes. IVFA was performed in all
patients within 8 weeks of the OCT angiography. Inclusion criteria
for diabetic patients included type 1 or type 2 diabetes with a clinical
examination by a retinal specialist confirming the presence of DR.
Two masked Boston Image Reading Center (BIRC) trained readers
reviewed both IVFA and OCT angiography images independently on
all patients with diabetes to identify microaneurysms and other retinal
abnormalities on both images. The size of the foveal avascular zone
(FAZ) and the perifoveal intercapillary area was measured in both
control and diabetic patients using the BIRC image analysis software.
The findings were then compared to determine the clinical utility of
OCT angiography compared with IVFA.
Results: The FAZ and perifoveal intercapillary area were enlarged
in patients with DR compared with controls. Ultra high speed swept
source OCT angiography was able to obtain detailed maps of the
retinal microvasculature and was able to identify the majority of
microaneurysms noted on fluorescein angiography with localization
of their exact retinal depth within a specific vascular plexus using
en face imaging. OCT angiography also revealed retinal vascular
abnormalities and microaneurysms not detected by IVFA.
Conclusions: Ultra high speed swept source OCT angiography
appears to be of significant utility in patients with DR, and is
completely non-invasive as compared with IVFA. It is able to detect
the majority of microaneursyms seen by IVFA, and is also able to
delineate other areas of retinal vascular abnormalities that were not
seen clinically or on IVFA. OCT angiography is also more accurate
for determining the FAZ and perifoveal intercapillary area. OCT
angiography may be of clinical utility in the evaluation and treatment
of diabetic patients.
Quantitative Analysis of Capillary Network Density in Diabetic
Retinopathy Using Optical Coherence Tomography with SplitSpectrum Amplitudinal Decorrelation Angiography
Steven Agemy1, Jessica Lee1, Patricia Garcia1, Yi-Sing Hsiao2, Toco
Y. Chui1, Richard B. Rosen1. 1Ophthalmology, New York Eye & Ear
Infirmary of Mount Sinai, New York, NY; 2OptoVue, Inc., Fremont,
CA.
Purpose: To quantitatively visualize retinal vascular flow in patients
with diabetic retinopathy using Optical Coherence Tomography
Angiography and a novel perfusion density mapping software.
Methods: OCT volumetric images of the macula were obtained
(3mm x 3mm and 6mm x 6mm) at 70 kHz A-scans per second using
the Optovue Avanti XR OCT system (Fremont, CA). Prototype
software employing the split-spectrum amplitudinal decorrelational
algorithm (SSADA) was used to construct SD-OCT angiograms of
the macular microvasculature. These images were then skeletonized
using Matlab software and processed to create topographic maps of
vascular density. Average perfusion density was also calculated for
the whole volumetric image.
Results: 18 eyes of 10 subjects with nonproliferative diabetic
retinopathy, 18 eyes of 9 subjects with proliferative diabetic
retinopathy, and 8 eyes of 4 control subjects were imaged. The
average perfusion density for the control group was 0.2477 ±0.0639
(3x3) and 0.2702 ±0.1006 (6x6). While the average perfusion density
for the NPDR group was significantly reduced at 0.2012 ±0.0694
(3x3) and 0.2474 ±0.1048 (6x6). The PDR group appeared futher
reduced at 0.1944 ±0.0692 (3x3) and 0.2402 ±0.1047 (6x6).
Conclusions: Topographic perfusion density mapping based upon
OCT angiography provides an easily interpretable quantitative
picture of retinal vascular flow. Using these novel perfusion density
maps, differences between normals and diabetic eyes at various
stages of retinopathy were easily recognizable. The ability to derive
quantitative values for average volumetric perfusion may also prove
useful for detecting progression and anticipating the need for more
aggressive interventions.
OCT angiography (left) versus IVFA (right) of the same eye with
NPDR.
Commercial Relationships: David A. Salz, None; Talisa de Carlo,
None; Mehreen Adhi, None; Eric Moult, None; Woo Jhon Choi,
None; Caroline R. Baumal, None; Andre J. Witkin, None; Jay
S. Duker, Carl Zeiss Meditech Inc. (F), EyeNetra (I), Hemera
Biosciences Inc. (I), Ophthotech Corp. (I), Optovue, Inc. (F); James
G. Fujimoto, Carl Zeiss Meditech, Inc. (P), Optivue, Inc. (I); Nadia
K. Waheed, None
SD-OCT angiography images (3mm x 3mm) of a 42 year old
male with proliferative diabetic retinopathy. Top: Enhanced image
Bottom: Topographic vessel density map with average perfusion
density of 0.1809 ± 0.0701 (skeletonized vessel count/total pixel
count)
SD-OCT angiography images (6mm x 6mm) of a 61 year old female
with nonproliferative diabetic retinopathy. Top: Enhanced image
Bottom: Topographic vessel density map with average perfusion
density of 0.2676 ± 0.1002 (skeletonized vessel count/total pixel
count)
Commercial Relationships: Steven Agemy, None; Jessica Lee,
None; Patricia Garcia, None; Yi-Sing Hsiao, OptoVue, Inc. (E);
Toco Y. Chui, None; Richard B. Rosen, OptoVue, Inc. (C)
Automated Quantification of Macular Ischemia Using Optical
Coherence Tomography Angiography in Diabetic Retinopathy
Yali Jia, Simon S. Gao, Thomas S. Hwang, Andreas K. Lauer, Steven
Bailey, Christina J. Flaxel, David J. Wilson, David Huang. Casey
Eye Institute, Oregon Health & Science University, Portland, OR.
Purpose: To quantify macular ischemia in diabetic retinopathy (DR)
using optical coherence tomography (OCT) angiography.
Methods: The macula of 7 eyes of healthy control subjects and 7
eyes with various levels of DR were imaged with a commercially
available 70 kHz OCT (RTVue XR, Optovue, Inc). Three dimensional
(3D) OCT angiography scans were acquired over 3×3 and 6×6 mm
regions by using 5 repeated B-scans at 216 raster positions, each
B-scan consisting of 216 A-scans. Flow was detected with the splitspectrum amplitude decorrelation angiography (SSADA) algorithm
and motion artifact was removed by 3D orthogonal registration and
merging of 2 scans. Retinal angiogram was created by projecting
the flow signal internal to the retinal pigment epithelium in en face
orientation. Parafoveal and perifoveal vessel density were defined as
the percentage of pixels with detectable flow in the respective regions
(Fig. 1B1). The threshold for detecting flow was set at 2.33 standard
deviations above the mean signal within the foveal avascular zone
(FAZ) (0.6 mm circle) of the control eyes. An automated algorithm
detected the area of macular low-perfusion, defined as the area with
flow signal below threshold within 5.5 mm from the foveal center
including the FAZ. (Fig. 1C).
Results: In 2 cases of DR, 6×6 mm scans were excluded due to
failed registration. 3×3 mm scans of these two cases were used
for the calculation of parafoveal vessel density. Compared to
normal controls, the parafoveal and perifoveal vessel density
were significantly reduced, low-perfusion area was significantly
greater in eyes with DR. Areas of low perfusion in OCT angiogram
corresponded to ischemic areas in fluorescein angiography.
Conclusions: OCT angiography can quantify retinal ischemia and
offers an objective and rigorous method of grading macular ischemia
in diabetic retinopathy.
Automated detection of nonperfusion in normal control (upper
panels) and diabetic retinopathy (lower panels). White dashed circle:
normal foveal avascular zone (0.6 mm diameter). Area between white
and blue dashed circles: parafoveal zone. Area between blue and
green dashed circles: perifoveal zone. Low-perfused areas (blue in C)
are detected by identifying flow signals lower than a set cutoff point.
Comparison of macular perfusion between normal controls and
diabetic retinopathy
Commercial Relationships: Yali Jia, Optovue, Inc (F), Optovue,
Inc (P); Simon S. Gao, None; Thomas S. Hwang, None; Andreas
K. Lauer, None; Steven Bailey, None; Christina J. Flaxel, None;
David J. Wilson, None; David Huang, Optovue, Inc (F), Optovue,
Inc (I), Optovue, Inc (P)
Support: DP3 DK104397, R01 EY024544, CTSA grant
(UL1TR000128), RPB, T32 EY23211
Changes in the Radial Peri-papillary Capillaries with aging: the
forgotten vascular bed in glaucoma pathogenesis
Tailoi Chan-Ling1, Samyoul Ahn1, Mark Koina2, Samuel J. Adamson1,
Marconi Barbosa3, Louise Baxter1, Frank Arfuso4, Anthony
Logaraj4, George Fatseas1. 1Retinal & Developmental Neurobiology
Laboratory, University of Sydney, Sydney, NSW, Australia; 2ACT
Pathology, Canberra Hospital, Garran, ACT, Australia; 3Diagnostics
for Eye Diseases Group, Eccles Institute for Neuroscience, JCSMR,
Australian National University, Canberra, ACT, Australia; 4School of
Anatomy, Physiology and Human Biology, The University of Western
Australia, Perth, WA, Australia.
Purpose: To examine if age-associated changes in the radial peripapillary capillaries (RPCs) constitute a significant component of
glaucoma pathogenesis.
Methods: 8 human fetal eyes, (8-40 weeks gestation (WG), 4 healthy
adult eyes (aged 17 to 33), and 12 aged human eyes (aged 40-86
years) were examined using multiple marker immunohistochemistry
(wholemounts and transverse sections) with antibodies against CD34,
CD39, GFAP, NG2, αSMA, S100β, and UEA-1 Lectin. Vascular
density, astrocyte and pericyte ensheathement, RPC distribution,
vascular branching patterns, and CD39 expression were examined
using confocal microscopy. RPC ultrastructure was examined using
TEM.
Results: RPCs were first evident between 18-22WG, in a narrow
rim surrounding the optic nerve head and expanded gradually until
birth. However, at birth RPC distribution was limited compared to
young adult. RPCs showed complete basal lamina and tight junctions
between adjacent VECs and were located within a 45um region from
the middle of the NFL and middle of GCL. Pericyte ensheathment
was examined ultrastucturally and using NG2 and αSMA IHC. The
frequency of pericyte ensheathement was 75% of that observed in the
capillaries of the inner retinal vascular bed. Remarkably, astrocytic
ensheathement examined both ultrastructurally and with antiGFAP and S100β IHC showed a very rare occurrence of astrocytic
ensheathement. RPC vascular density index (33% at 32 years
compared to 20% at 80 years) and distribution decreased markedly
with age. Using threshold intensity analysis, we demonstrated that
CD39 expression decreased with age. Importantly, many RPC
segments became discontinuous and rheology and blood flow was
likely impeded in the most aged specimens.
Conclusions: Our novel observation that RPCs have a constitutive
paucity of astrocytic ensheathment has implications for their ability
to auto-regulate blood flow in response to neuronal activity. Taken
together, our data leads us to suggest that the lack of astrocyticvascular interactions on the RPCs and their marked atrophy in aging
compromises their ability to maintain homeostasis and support
neuronal function in ‘physiological aging’ and contribute to glaucoma
pathogenesis. Our findings show a correlation between decreasing
RPC vascular perfusion with aging but does not demonstrate
causality between vascular changes and GC loss.
Commercial Relationships: Tailoi Chan-Ling, None; Samyoul
Ahn, None; Mark Koina, None; Samuel J. Adamson, None;
Marconi Barbosa, None; Louise Baxter, None; Frank Arfuso,
None; Anthony Logaraj, None; George Fatseas, None
Support: National Health and Medical Research Council of Australia
(Nos. 1005730 & 571100), the Baxter Charitable Foundation, the
Alma Hazel Eddy Trust and the Rebecca L. Cooper Medical Research
Foundation
Swept-Source OCT Angiography (OCTA) of Subjects with
Retinal Vein Occlusions
Sun Young Lee1, Douglas Matsunaga1, Jack Yi1, Mary K. Durbin2,
Carmen A. Puliafito1, Amir H. Kashani1. 1Ophthalmology, USC Eye
Institute, Los Angeles, CA; 2Advanced Development, Carl Zeiss
Meditec, Inc, Dublin, CA.
Purpose: To assess the retinal vasculature in patients with retinal
vein occlusion (RVO) using OCTA.
Methods: Cross-sectional, observational study of subjects with RVO.
All subjects underwent complete ophthalmic examination and had
a confirmed diagnosis of RVO. Data was acquired using a Cirrus
high-speed 1050 nm SS-OCT prototype system (100,000 kHz) and a
840 nm SD-OCT prototype system (67,500 kHz)(Carl Zeiss Meditec,
Dublin, CA) on 3x3mm sections of the macula and compared to
conventional fluorescein angiography (FA). Retinal vasculature was
assessed in three retinal slabs consisting of the superficial retina
(inner limiting membrane to superficial inner plexiform layer),
middle retina (deep inner plexiform layer to superficial outer nuclear
layer) and outer retina (deep outer nuclear layer to external limiting
membrane). The vasculature was reconstructed using a phase
and intensity contrast based algorithm and visualized en face for
comparison with fluorescein angiograms (FA).
Results: OCTA in subjects with RVO (N=5) clearly demonstrates
areas of non-perfusion, tortuous or dilated vessels and abnormal
capillary morphology. These findings are qualitatively consistent with
findings on fluorescein angiography (FA). Dilated and anomalous
capillaries were more clearly defined in OCTA than in most frames
of fluorescein angiograms. This is particularly evident in late FA
images with even mild leakage that tended to washout surrounding
capillaries. There were no findings on OCTA that correlated with
areas of FA leakage in any case. OCTA demonstrated clear areas of
capillary non-perfusion that correlated well with FA findings. Using
OCTA, the capillary non-perfusion could be localized to both inner
and middle retinal layers in all cases.
Conclusions: OCTA provides high-resolution and non-invasive
angiograms that correlate well with most findings on fluorescein
angiography, except dye leakage. OCTA shows that capillary nonperfusion in RVO involves both inner and middle retinal layers.
Commercial Relationships: Sun Young Lee, Carl Zeiss Meditec
(F); Douglas Matsunaga, Carl Zeiss Meditec (F); Jack Yi, Carl
Zeiss Meditec (F); Mary K. Durbin, Carl Zeiss Meditec (E);
Carmen A. Puliafito, Carl Zeiss Meditec (F); Amir H. Kashani,
Carl Zeiss Meditec (F)
Support: Unrestricted grant from Research to Prevent Blindness,
New York, NY 10022
Analysis of vessels functionality in retinal vein occlusion treated
with intravitreal ranibizumab
Federico Corvi1, Carlo La Spina1, Lucia Benatti1, Lea Querques1,
Rosangela Lattanzio1, Francesco Bandello1, Giuseppe Querques1, 2.
1
Ophthalmology, IRCCS, Vita-Salute San Raffale University, Milan,
Italy; 2University Paris Est Creteil, Centre Hospitalier Intercommunal
de Creteil, Creteil, France.
Purpose: To investigate the effects of intravitreal injection of
ranibizumab on retinal vessels functionality in patients with retinal
vein occlusion (RVO).
Methods: Consecutive treatment-naïve patients with macular
edema secondary to RVO were enrolled in this prospective study.
All patients underwent a complete ophthalmic evaluation, including
optical coherence tomography and dynamic and static retinal
vessel analysis using the Dynamic Vessels Analyzer (DVA) before
(baseline),1 week and 1 month after administration of intravitreal
ranibizumab. Investigation of RVO patients were compared to ageand sex-matched control subjects.
Results: We included a total of 11 eyes of 11 patients with ME
secondary to RVO (10 men; mean age 56.8±13.3 years). In RVO
patients, dynamic analysis showed a significant increase of mean
venous dilation form +2.46±1.03% at baseline to +3.96±1.3% at
1 week (p=0.001). At 1-week mean maximum venous and arterial
dilations did not differ from control subjects. Static analysis
showed a mean overall significant decrease of CRAE and CRVE
from baseline to 1 week (from 174.8±22.5MU to 167.2±26.7MU
[p=0.04], and from 228.4±20.7MU to 217.3±22.8 [p=0.0002]).
Mean CRAE in healthy control subjects was 175.9±10.45MU, not
significantly different from baseline, week-1 and month-1 of RVO
eyes. Conversely, mean CRVE was 195.5±9.91 MU in healthy
control subjects, significantly different from baseline, week 1 and
month 1 of RVO eyes. By considering only each single occluded
quadrant in the 11 RVO eyes (for a total of 29 quadrants), mean
CRAE significantly decreased at 1 week (from 102.3±19.7MU to
192.7±21.6MU [p=0.003]). Mean CRVE for occluded quadrants
significantly decreased at both 1 week and 1 month (from
136.4±24.5MU to 123.5±24.7MU [p<0.0001]). By considering
only the non-occluded quadrants (only BRVO, for a total of 15
quadrants), mean CRAE did not change from baseline to week-1 and
month-1 (from 101.8±14.2MU to 100±15.6MU and to 103.2±15MU
[p=0.9]). Similarly, mean CRVE for non-occluded quadrants did not
change from baseline to week-1 and to month-1 (126.1±12MU to
123.2±13.4MU and to 124.1±10.9MU [p=0.9]).
Conclusions: Using DVA in patients with RVO we found that
intravitreal ranibizumab increased veins dilation (dynamic analysis),
and had a vasoconstrictive effect on both arteries and veins,
especially in the occluded quadrants (static analysis).
Commercial Relationships: Federico Corvi, None; Carlo
La Spina, None; Lucia Benatti, None; Lea Querques, None;
Rosangela Lattanzio, None; Francesco Bandello, ALIMERA (C),
ALLERGAN (C), BAYER (C), FARMILA-THEA (C), GENETECH
(C), HOFFMAN-LAROCHE (C), NOVAGALI PHARMA
(C), NOVARTIS (C), SANOFI-AVENTIS (C), SCHERINGPHARMA (C); Giuseppe Querques, ALCON (C), ALIMERA
(C), ALLERGAN (C), BAUSCH AND LOMB (C), BAYER (C),
NOVARTIS (C), OPHTHOTECH (C)
Evaluation of choroidal and retinal vasculature network in
patients with retinitis pigmentosa using optical microangiography
Kasra Attaran-Rezaei1, Qinqin Zhang2, Jennifer R. Chao1, Yanping
Huang2, Ruikang K. Wang2. 1Ophthalmology, University Of
Washington, Seattle, WA; 2bioengineering, university Of Washington,
Seattle, WA.
Purpose: Retinitis pigmentosa (RP) is a heterogeneous group of
hereditary retinal diseases that result in progressive loss of rod and
cone photoreceptors. The ocular blood circulation has been shown to
be altered in RP in many experimental and clinical studies. Optical
coherence tomography (OCT) based microangiography (OMAG) was
recently illustrated for the functional imaging of the microvascular
network within the tissue beds. In this study we evaluated the retinal
and choroidal microvascular architecture in RP patients using
OMAG.
Methods: Eight patients (sixteen eyes) with Retinitis Pigmentosa
underwent OMAG. OMAG was performed by Zeiss spectral domain
OCT-angiography prototype using a 6 mm X 6 mm field of view
around macular region. The resulting retinal image was segmented
into two layers: the inner retinal layer from the ganglion cell layer
to the inner plexiform layer, the deeper retinal layer from the inner
nuclear layer to the external limiting membrane. The choroidal
image was segmented into choriocapillaris and choroidal layers. The
vascular distribution in each layer was depicted as an enface image.
Results: In all eyes with RP imaged by OMAG, abnormal
microvasculature was detected in both the deeper retinal vasculature
layer (from the inner nuclear layer to the external limiting membrane)
and choroidal vasculature. A representative result from one patient
(left eye) is provided in the figure attached. The OMAG results
correlated very well with visual field testing and with reduced
choroidal-retina thickness measured by Zeiss SD-OCT-angiography
prototype in RP patients.
Conclusions: OMAG is a new imaging technique that can evaluate
the microvascular network changes in the retina and choroid. OMAG
imaging provided detailed, depth-resolved information about the
microvasculature changes in Retinitis Pigmentosa patients. OMAG
shows loss of normal vessel architecture in the choriocapillaris,
choroid and deeper retinal vascular layer. These images corresponded
well with published clinical and histological findings.
Commercial Relationships: Kasra Attaran-Rezaei, None; Qinqin
Zhang, None; Jennifer R. Chao, None; Yanping Huang, None;
Ruikang K. Wang, Carl Zeiss Meditec Inc (P)
Support: R01EY024158. P30-EY001730. Research to Prevent
Blindness Unrestricted Grant.
Quantitative optical coherence tomography angiography of the
choriocapillaris in central serous chorioretinopathy
Scott M. McClintic, Yali Jia, Steven Bailey, Simon S. Gao, David
Huang. Ophthalmology, Casey Eye Institute, Oregon Health &
Science University, Portland, OR.
Purpose: To determine if optical coherence tomography (OCT)
angiography can detect abnormal choriocapillaris blood flow in
central serous chorioretinopathy (CSC) compared to a group of
normal controls.
Methods: Five eyes of 5 study participants with CSC (1 acute,
4 chronic) and 5 eyes of 5 normal controls were scanned with a
spectral domain OCT (RTVue XR; 70 kHz scanning speed). Macular
angiograms (3x3 mm, 6x6 mm) were obtained using the spitspectrum amplitude decorrelation angiography (SSADA) algorithm,
which detects areas of flow in otherwise static tissue. The volumetric
angiogram was segmented into inner retinal, outer retinal, and
choriocapillaris (10 microns below BM). Color-coded OCT cross
sections allowed for representation of both flow and structure (purple
= inner retina flow, red = outer retina flow). Vessel density (VD)
of the choriocapillaris was the percentage of pixels with detectable
flow in the segmented en face choriocapillaris angiogram. Statistical
analysis included calculation of mean ± standard deviation and the
Mann Whitney U test for group comparison.
Results: Mean choriocapillaris VD in the 3x3 mm scan was
significantly reduced in CSC subjects (93.93% ± 3.99%) compared
to normals (98.64% ± 0.61%), P<0.009. En face OCT angiograms in
normal participants revealed relatively homogenous choriocapillaris
flow. CSC subjects had patchy areas of decreased flow, which
correlated to regions of hypo-cyanescence on indocyanine green
angiography (ICGA). Hyperfluorescent staining and leakage on
fluorescein angiography (FA) correlated to most, but not all, areas of
reduced choriocapillaris flow.
Conclusions: OCT angiography detected reduced choriocapillaris
vessel density in CSC subjects compared to normal controls. Patchy
areas of reduced flow were visible on en face choriocapillaris
angiograms. Further study of the choriocapillaris with OCT
angiography may improve understanding of CSC pathogenesis.
Figure 1. A-E: Chronic CSC participant, including midphase
FA (A) and ICGA (B), cross section OCT angiogram at level of
green line (C), inner retina OCT angiogram (D), choriocapillaris
OCT angiogram (E). F: Choriocapillaris OCT angiogram for
normal participant. White box corresponds to sample area of OCT
angiogram.
Commercial Relationships: Scott M. McClintic, None; Yali Jia,
Optovue, Inc. (F), Optovue, Inc. (P); Steven Bailey, None; Simon
S. Gao, None; David Huang, Carl Zeiss Meditec, Inc. (P), Optovue,
Inc. (F), Optovue, Inc. (I), Optovue, Inc. (P)
Support: R01 EY023285, R01 EY024544, DP3 DK104397, RPB,
CTSA grant (UL1TR000128), NIH T32 EY23211
Optical Coherence Tomography Angiography of chronic central
serous chorioretinopathy in elderly patients
Mariachiara Morara2, Chiara Veronese2, Martina Melucci1, Filippo
Tassi1, Nicole Balducci2, Antonio Ciardella2. 1University of Bologna
- Policlinico S.Orsola-Malpighi, Bologna, Italy; 2Policlinico S.Orsola
- Malpighi, Bologna, Italy.
Purpose: To noninvasively describe the features of retinal and
choroidal vascular changes in elderly patients with central serous
chorioretinopathy (CSC) with Optical Coherence Tomography
Angiography (OCTA).
Methods: Cross-sectional, observational study of 10 patients
between 55 and 64 years of age with central serous chorioretinopathy.
OCTA was performed on 3x3 and 6x6 mm area centered on the
fovea. The 3D angiography was segmented in 4 layers: superficial
and deep (to show retinal vasculature), outer retina (to identify
Choroidal neovascularization) and chorio-capillary.
Results: Serous retinal detachment was clearly visualized by
OCTA in 10 eyes of 10 patients with CSC. A thin Pigment epithelial
detachment (PED), that is difficult to identify with conventional
images modalities, was identified in 10 eyes of 10 patients. In 4
patients OCTA detected a distinct choroidal neovascular pattern in
chorio-capillary.
Conclusions: According to PED morphology in CSC, OCTA allows
to detect associated CNV, previously undiagnosed by conventional
imaging modalities.
Commercial Relationships: Mariachiara Morara, None; Chiara
Veronese, None; Martina Melucci, None; Filippo Tassi, None;
Nicole Balducci, None; Antonio Ciardella, None
Optical coherence tomography angiography (OCTA) detection
of choroidal neovascularization (CNV) in chronic central serous
chorioretinopathy (CSCR)
Marco A. Bonini1, 2, Talisa de Carlo2, 3, Daniela Ferrara2, Caroline
Baumal2, Andre J. Witkin2, Elias Reichel2, Jay S. Duker2, Nadia K.
Waheed2. 1Ministry of Education, Brasília, Brazil; 2Retina, New
England Eye Center, Boston, MA; 3Massachussets Insttitute of
Technology, Boston, MA.
Purpose: To evaluate the sensitivity of spectral-domain OCTA in
detecting CNV associated with pigment epithelial detachment (PED)
in chronic CSCR
Methods: Eyes that were previously diagnosed with chronic
CSCR and that were receiving multimodal imaging for suspicion
of CNV were prospectively recruited to receive additional ancillary
imaging test using the prototype AngioVue OCTA software on a
commercially-available spectral-domain OCT (SD-OCT) device to
generate en-face images (OCT angiograms) (Optovue, Inc, Fremont,
CA). Orthogonal registration and merging of two consecutive image
sets is used to obtain 3x3mm and 6x6mm OCT angiograms. The
OCTA software was used to delineate a region of interest with an
inner boundary at the level of the outer aspect of the outer plexiform
layer and an outer boundary at the level of Bruch’s membrane.
An “artifact removal” function within the software was utilized
to subtract retinal vessel shadowing from the OCT angiogram.
Qualitative analysis was performed based on OCTA findings of
vascular flow representing CNV and CNV appearance. Crosssectional OCT b-scans were used to determine CNV location relative
to the retinal pigment epithelium and Bruch’s membrane.
Results: Twenty-seven eyes of 23 consecutive CSCR patients
demonstrating chronic CSCR and associated PED were enrolled in
this study. Dye based angiography showed clear evidence of CNV
in all eyes. From the eight eyes with confirmed CNV, five eyes had
well-circumscribed vessels in the CNV area and three eyes showed
poorly-circumscribed vessels on OCTA. Manual displacement of the
outer boundary segmentation line down to the choriocapillaris level
on correlating OCT b-scans was performed and allowed delineation
of CNV boundaries in 6/8 eyes (75%) and feeder vessel in 2/8 (25%)
eyes with CNV.
Conclusions: We demonstrated agreement between non-invasive
OCTA and dye-based angiography in detecting CNV associated
with chronic CSCR. This study suggests that OCTA may be a viable
alternative to dye based angiography in the diagnosis of CNV in
patients with chronic CSCR. Since OCTA is completely non-invasive
it could at least be considered a first step in establishing the diagnosis.
Future studies with larger sample size are needed to improve
our understanding of this diagnostic method, and provide further
information to validate this imaging technique in clinical practice.
Commercial Relationships: Marco A. Bonini, None; Talisa de
Carlo, None; Daniela Ferrara, None; Caroline Baumal, None;
Andre J. Witkin, None; Elias Reichel, None; Jay S. Duker, Carl
Zeiss Meditec (C), Optovue (C); Nadia K. Waheed, None
DYNAMIC AND STATIC VESSELS ANALYSIS IN
PATHOLOGIC MYOPIA
Lucia Benatti1, Carlo La Spina1, Federico Corvi1, Giuseppe
Querques1, 2, Francesco Bandello1. 1Ophthalmology, University VitaSalute San Raffaele, Milan, Italy; 2Ophthalmology, Centre Hospitalier
Intercommunal de Creteil, Creteil, France.
Purpose: To investigate retinal vascular anatomy and function in
eyes with pathologic myopia by Dynamic Vessel Analyzer (DVA,
Imedos, Jena, Germany) in order to explain why these eyes are
resistant to diabetes-related changes.
Methods: A total of 20 patients with pathological myopia (20 eyes)
were included in the study and compared with 20 eyes of age and
sex matched healthy control subjects. Complete ophthalmologic
examination, dynamic and static retinal vessels analysis were
performed on all participants.
Results: The dynamic analysis of eyes stimulated by flickering light
highlighted a mean arterial dilation of 2.44±1.59% in myopic eyes
and 3.28±1.76% in healthy eyes (p=0.189). Mean venous dilation
was 3.45±1.82% and 4.45±2.72%, respectively (p=0.409). In myopic
patients, the static retinal analysis showed a mean Central Retinal
Artery Equivalent (CRAE) of 171.6±24.3, a mean Central Retinal
Vein Equivalent (CRVE) of 199.5±27.73 and a mean arteriovenous
ratio (AVR) of 0.86± 0.01. In control subjects mean CRAE was
190.3±11.93 and mean CRVE 215.7±13.30 (both p = 0.0031 vs
respective measurements in myopia) and mean AVR was 0.88±0.04
(p=0.913). An inverse relationship between axial length and CRAE
and CRVE (r2 = 0.18, p=0.0051 and r2 = 0.18, p=0.0165, respectively)
was evidenced by the linear regression analysis.
Conclusions: Static and dynamic vessels analysis showed that
myopic eyes have retinal posterior pole vessels with reduced
diameter, but with normal function. The vascular network at the
posterior pole does not appear to play a key role in determining
chronic state of ischemia involved in the relative protection from
diabetic retinopathy.
Commercial Relationships: Lucia Benatti, None; Carlo La Spina,
None; Federico Corvi, None; Giuseppe Querques, ALIMERA
(C), ALLERGAN (C), BAUSCH AND LOMB (C), BAYER (C),
NOVARTIS (C), OPHTHOTECH (C); Francesco Bandello, ALCON
(C), ALIMERA (C), ALLERGAN (C), BAUSCH AND LOMB (C),
BAYER (C), FARMILA-THEA (C), GENETECH (C), NOVARTIS
(C), PFIZER (C), THROMBOGENICS (C)
Optical Coherence Tomography Angiography of Pigment
Epithelial Detachment
Chiara Veronese1, Mariachiara Morara1, Martina Melucci2, Filippo
Tassi2, Nicole Balducci1, Antonio Ciardella1. 1Policlinico S.OrsolaMalpighi, Bologna, Italy; 2University of Bologna, Bologna, Italy.
Purpose: To noninvasively describe the spectrum of pigment
epithelial detachments (PEDs) occurring mainly in age-related
macular degeneration (AMD), central serous chorioretinopathy
(CSC), inflammatory and iatrogenic retinal disorders with Optical
Coherence Tomography Angiography (OCTA).
Methods: Observational, cross-sectional study of 34 patients (19
men and 15 women), ranged in age from 50 to 70 years (mean
age 63.2 years) with drusenoid, serous, vascularized or mixed
PEDs. The instrument used for the OCT images was based on the
RTVue XR Avanti (Optovue Inc) and was used to obtain splitspectrum amplitude decorrelation angiography images. OCTA was
performed on 3x3 mm and 6x6 mm area centered on the fovea. The
3D angiography was segmented in 4 layers: superficial and deep
(to show retinal vasculature), outer retina (to identify Choroidal
neovascularization) and chorio-capillary. En face maximum
projection was used to obtain 2-dimensional angiograms from the 4
layers.
Results: En face OCT angiograms of PEDs showed sizes and
locations that were confirmed by fluorescein angiography (FA).
OCTA of 34 eyes detected 34 PEDs : vascularized in 21 eyes
(61.8%), serous in 7 eyes (20.6%), drusenoid in 4 eyes (11.8%),
infiammatory in 1 eye (2.9%) and iatrogenic in 1 eye (2.9%).
Conclusions: OCTA provides depth-resolved information and
detailed images of PEDs and may offer noninvasive differentiation
between various kinds of PEDs.
Commercial Relationships: Chiara Veronese, None; Mariachiara
Morara, None; Martina Melucci, None; Filippo Tassi, None;
Nicole Balducci, None; Antonio Ciardella, None
The dark atrophy: an angio-OCT study
Alessandra Acquistapace, Marco Pellegrini, Marta Oldani, Matteo
G. Cereda, Andrea Giani, Vittoria Ravera, Giovanni Staurenghi.
Sacco Hospital Eye Clinic, Department of Clinical Science, San
Fermo della Battaglia, Italy.
Purpose: to evaluate the status of choriocapillaris and choroidal
structures using angio-OCT in patients affected by macular atrophy
secondary to age-related macular degeneration (GA) and Stargardt
disease (STGDT).
Methods: Retrospective study. All the patients underwent a
complete ophthalmological examination including blue and green
autofluorescence (B-FAF, G-FAF), fluorescein (FA) and indocyanine
green (ICGA) angiography, enhanced depth imaging spectral
domain optical coherence tomography (SD-EDI OCT) (HRA+OCT
Spectralis, Heidelberg Engineering, Heidelberg, Germany) and angioOCT using AngloVue technologies (Optovue, in).
Results: 20 eyes of 10 patients affected by GA and 20 eyes of
10 patients affected by STGDT were included in the study. Mean
age was 53,3 for STGDT patients and 76,88 for GA. Atrophy was
hypofluorescent in both B-FAF and G-FAF in all the cases. In the
early frames FA displayed hyperfluorescence in the atrophic area in
80% of GA and 0% of STGDT patients whereas dark choroid was
present in 0% of patients and 90% respectively. ICGA showed, in late
frames, hypocyanescence in 95% of STGTD and 16% of GA patients
and isocyanescence in 84% of GA patients. At Angio-OCT imaging
it was possible to detect a global rarefaction of choroidal layers in
94% of GA patients with no selective loss of choriocapillaris while in
STGDT patients loss of choriocapillaris was typically antecedent to
the loss of the other layers.
Conclusions: SSADA analysis in STGDT patients displayed a
selective involvement of choriocapillaris compared to GA population.
This finding results in agreement with possible expression of ABCA4
at this level and with previous ICGA studies.
Commercial Relationships: Alessandra Acquistapace, None;
Marco Pellegrini, Bayer (S); Marta Oldani, None; Matteo G.
Cereda, None; Andrea Giani, Novartis Advisory Board (C);
Vittoria Ravera, None; Giovanni Staurenghi, Alcon Laboratories,
Inc (C), Allergan, Inc (C), Bayer (C), Boehringer (C), Genentech
(C), GlaxoSmithKline (C), Heidelberg Engineering (C), Novartis
Pharmaceuticals corporation (C), Novartis Pharmaceuticals
corporation (S), Optos, Inc (C), Optos, Inc (C), Optovue (S), Roche
(C), Zeiss (C), Zeiss (S)
OCT angiography provides insights into choroideremia pathology
Nieraj Jain, Yali Jia, Simon S. Gao, Michael J. Gale, David Huang,
Richard G. Weleber, Mark E. Pennesi. Casey Eye Institute, Oregon
Health & Science University, Portland, OR.
Purpose: To explore choriocapillaris (CC) alterations in
choroideremia (CHM) with optical coherence tomography (OCT)
angiography and to correlate these findings with deficits in retinal
structure and visual function.
Methods: This is a prospective study of subjects with CHM, CHM
carrier state, and normal controls. Subjects underwent imaging with
70 KHz spectral OCT (RTVue-XR) in 3x3mm and 6x6mm macular
areas using the split-spectrum amplitude-decorrelation angiography
approach. Segmentation of the CC with en face maximal projection
provided a 2 dimensional angiogram. Adaptive optics (AO) images
were analyzed with custom cone-counting software to provide a
cone density map. Additional structural imaging included fundus
autofluorescence and color fundus photography. Visual function
was assessed through best corrected visual acuity (BCVA) and
microperimetry using a custom 101 point testing grid. Image
registration was performed to evaluate the spatial relationship
between CC density and structural and functional metrics of the
retina and retinal pigment epithelium (RPE).
Results: 6 subjects with CHM, 2 carriers, and 7 controls were
enrolled. BCVA ranged from 20/20 to 20/30 in affected males and
20/20 to 20/60 in carrier females. Mean age amongst affected males
and carriers was 44.5±15.1 years. OCT angiography demonstrated
regions nearly devoid of CC amongst affected males, and patchy
areas of reduced CC density amongst carriers. In affected males,
there was strong spatial correlation between CC density, retinal and
RPE anatomy, and visual function. Subtle discrepancies were noted
at transition zones. Amongst carrier females, there were greater
discrepancies. In some carrier eyes, reduced CC density and visual
function coincided in areas with apparently normal retinal structure.
Conclusions: OCT angiography permits in vivo CC assessment not
previously possible, and demonstrates a spectrum of CC pathology in
CHM. Structure-function correlations provide new insights into the
disease pathology. This technology offers promising new endpoints
for upcoming clinical trials in CHM.
Registered en face OCT images in a choroideremia patient
demonstrate (A) segmented photoreceptor reflectance; (B) total
reflectance depicting the region of intact RPE (central dark region);
and (C) OCT angiogram with regional variability in choriocapillaris
flow (high flow represented by bright gold; no flow represented by
black).
Commercial Relationships: Nieraj Jain, None; Yali Jia, Opovue,
Inc. (F), Opovue, Inc. (P); Simon S. Gao, None; Michael J. Gale,
None; David Huang, Carl Zeiss Meditec, Inc. (P), Optovue, Inc.
(F), Optovue, Inc. (I), Optovue, Inc. (P); Richard G. Weleber, U.S.
patent no. 8657446 (P); Mark E. Pennesi, Sucampo Pharmaceuticals
(C)
Support: NIH grants R01 EY023285, R01 EY024544, 1K08
EY0231186-01 (Pennesi), DP3 DK104397, T32 EY23211,
P30EY010572 core grant, CTSA grant (UL1TR000128); FFB grants
CDA CF-CL-0614-0647-OHSU (Jain), Enhanced CDA (Pennesi);
RPB CDA (Pennesi)
Evaluation of Age-related Macular Degeneration and Polypoidal
Choroidal Vasculopathy using OCT-based Microangiography
Ruikang K. Wang1, Qinqin Zhang1, Cecilia Lee2, Yanping Huang1,
Kasra Attaran Rezaei2, Richard Munsen2, Jennifer R. Chao2, James L.
Kinyoun2. 1Bioengineering, University of Washington, Seattle, WA;
2
Ophthalmology, University of Washington, Seattle, WA.
Purpose: To assess the feasibility and proficiency of OCT-based
microangiography (OMAG) in the detection and visualization of
vascular involvement at different stages of age-related macular
degeneration (AMD).
Methods: Twenty patients were recruited, including early stage
AMD, geographic atrophy (GA), neovascular AMD and polypoidal
choroidal vasculopathy (PCV). Patients were scanned by a Zeiss
OCT-angiography prototype with motion tracking. OCT scans
centered on the fovea were captured to generate the OMAG images.
OMAG images were segmented into 4 layers including the inner
(GCL->IPL), deeper (INL->photoreceptor layer) retinal layers,
choriocapillaris, and deeper choroidal layer. Enface images were
used to represent angiograms at different layers (coded with different
colors for visual purposes).
Results: Enface OMAG images showed overall good agreement
with fluorescein angiography (FA). OMAG gave more detailed
visualization of vascular networks that were less affected by
subretinal hemorrhages. In early stage AMD, small drusen were
observed, but retinal vessels seemed the same as for normal subjects.
For patients with GA, abnormalities of the RPE layer resulted in
the ability to easily observe the choriocapillaris and large choroidal
vessels. Feeding and draining vessels were identified in neovascular
AMD and PCV. Choroidal neovascularization (CNV) was more
demarcated by OMAG as compared to FA due to the late leakage of
fluorescein dye which obscures the CNV.
Conclusions: OMAG provides depth-resolved and detailed vascular
images of AMD. In most cases, proper segmentation is the key to
identifying the location of abnormal vessels. Our ongoing studies
with OMAG will standardize quantification of the retinal and
choroidal vascular layers during the progression of AMD as well as
following treatment, particularly with anti-VEGF agents.
PCV of serosanguineous PEDs from 72-year old patient. (A, B)
are the early and late phase FA images; (C) The color OMAG
image from retina to choroid overlying the FA image; (D) The
magnified FA image corresponding to the OMAG images; (E) The
vasculature within inner (red) and deeper (green) retinal layer. (F)
The vasculature from RPE to choroid layers. (G) The cross-sectional
structure image at the position marked by the white line in (E, F)
overlying flow image; color coding scheme is shown.
Commercial Relationships: Ruikang K. Wang, Carl Zeiss Meditec
Inc (F), Carl Zeiss Meditec Inc (P); Qinqin Zhang, Carl Zeiss
Meditec (F); Cecilia Lee, None; Yanping Huang, None; Kasra
Attaran Rezaei, None; Richard Munsen, None; Jennifer R. Chao,
None; James L. Kinyoun, None
Prevent Blindness.
OCT Angiography (OCTA) of Macular Telangiectasia Type 2
John E. Legarreta1, Andrew D. Legarreta1, Mariana R. Thorell1,
Qinqin Zhang2, Giovanni Gregori1, Douglas Matsunaga3, Amir
H. Kashani3, Ruikang K. Wang2, Carmen A. Puliafito3, Philip
J. Rosenfeld1. 1Ophthalmology, Bascom Palmer Eye Institute,
University of Miami Miller School of Medicine, Miami, FL;
2
Bioengineering, University of Washington, Seattle, WA; 3USC Eye
Center, Keck School of Medicine, University of Southern California,
Los Angeles, CA.
Purpose: To evaluate the microvasculature of the central macula in
eyes with macular telangiectasis type 2 (MacTel2)
using swept-source (SS) and spectral-domain (SD) optical coherence
tomography (OCT) angiography.
Methods: Subjects were enrolled in a prospective, observational
study and evaluated using a high-speed 1050 nm
SS-OCT prototype system (100,000 kHz) and a 840 nm SD-OCT
prototype system (68,000 kHz). SS-OCT
angiography was performed using a 3X3mm raster scan pattern
centered on the fovea. In the transverse scanning direction, a single
B-scan was comprised of 300 A-scans. Four consecutive B-scans
were performed at each fixed position before proceeding to the next
transverse position on the retina. A total of 300 B-scan positions
located 10 μm apart over a 3 mm distance were sampled. SD-OCT
angiography was performed using a 3X3mm and 6X6mm raster
scan pattern. In the 3X3 raster scan, four consecutive B-scans, each
comprised of 245 A-scans, were performed in the transverse scanning
direction. In the 6X6 raster scan pattern, the transverse scanning
direction was comprised of two consecutive B-scans, each of 350
A-scans. A total of 245 B-scan positions were located 12.4 μm apart
over the 3 mm distance, and a total of 350 B-scan positions were
located 17.1 μm apart over the 6 mm distance. Algorithms segmented
the retina into three layers; an inner retinal layer, a middle retinal
layer, and an outer retinal layer. The vascular distribution in each
layer was depicted as an en face image. En face OCT angiographic
images were compared to early and late phase fluorescein
angiography (FA) images.
Results: OCT angiography imaging was performed on 35 subjects
(67 eyes) with MacTel2. In all MacTel2 eyes, an abnormal
microvasculature was detected. In early stages of MacTel2, these
abnormal vessels appeared to reside predominantly in the middle
retinal layers or deep capillary plexus. The earliest manifestations
included dilation and truncation of the microvasculature in the
temporal juxtafoveal location. As the disease progressed, all retinal
layers became involved and demonstrated abnormal vascular
patterns.
Conclusions: OCT angiography can be performed with both SS-OCT
and SD-OCT instruments and provides rapid, non-invasive, highresolution, depth-resolved images comparable to or even better than
conventional fluorescein
angiography in eyes with MacTel2.
Commercial Relationships: John E. Legarreta, Carl Zeiss
Meditec, Inc (F); Andrew D. Legarreta, Carl Zeiss Meditec, Inc
(F); Mariana R. Thorell, Carl Zeiss Meditec, Inc (F); Qinqin
Zhang, Carl Zeiss Meditec, Inc (F); Giovanni Gregori, Carl Zeiss
Meditec, Inc (F), Carl Zeiss Meditec, Inc (P); Douglas Matsunaga,
Carl Zeiss Meditec, Inc (F); Amir H. Kashani, Carl Zeiss Meditec,
Inc (F); Ruikang K. Wang, Carl Zeiss Meditec, Inc (F); Carmen
A. Puliafito, Carl Zeiss Meditec, Inc (F); Philip J. Rosenfeld, Carl
Zeiss Meditec, Inc (F)
Support: Research Support from Carl Zeiss Meditec, Inc
Optical Microangiography Imaging in a Patient with Retinal
Vasculopathy from Susac’s Syndrome
Raghu C. Mudumbai1, Qinqin Zhang1, 2, Chieh-Li Chen1, 2, Yanping
Huang1, 2, Ruikang K. Wang1, 2. 1Ophthalmology, University of
Washington Eye Institute, Seattle, WA; 2Bioengineering, University
of Washington Medical Center, Seattle, WA.
Purpose: To investigate the retinal perfusion differences in a patent
found to have a focal retinal vasculopathy from Susac’s syndrome
using optical microangiography (OMAG).
Methods: A 22 year old woman was diagnosed with Susac’s
syndrome based on a triad of intracranial lesions including
involvement of the corpus callosum on MRI, hearing loss, and vision
loss. Ophthalmic examination showing 2 cotton wool spots along the
inferotemporal arcade OS. HVF indicated an area of a deep defect
superiorly OS that was symptomatic to the patient. Fluoroscein
angiography (FA) indicated a corresponding filling defect in the
arterioles without signs of pseudo-emboli consistent with Susac’s
syndrome.
The patient’s retina was scanned with a 67 kHz Cirrus 5000 HD-OCT
based angiography prototype system with motion tracking (Zeiss,
Dublin, CA) with a 6 mm x 6 mm field of view to provide 3D OMAG
retinal microvascular maps. Using proprietary semi-automatic
segmentation software, retinal OMAG images were segmented into
3 layers including Inner retinal layer (ganglion cell + inner plexiform
layer), middle retinal layer (inner nuclear + outer plexiform layer)
and outer retinal layer (outer nuclear + photoreceptor layer). Enface
maximum projection was used to represent angiograms at different
layers (coded with different colors for visual purpose).
Results: Results showed strong correlation between OMAG and the
area of retinal non-perfusion identified by FA. Quantitative analysis
of the retinal vessel density (the ratio of retinal vessel area over the
area evaluated) indicated a sharp drop in the retinal perfusion of the
effected zone compared to the surrounding retina. Comparison of the
retinal hemifields also indicated a significant difference between the
effected zone and the corresponding area in the opposite hemifield.
Cross sectional analysis of retinal perfusion zones indicated
diffuse loss of both the inner and outer retinal layers, slightly more
pronounced in the middle retinal vessels. See figure.
Conclusions: OMAG is able to provide additional information
that cannot be obtained with FA that quantifies the severity of
non-perfusion as well as cross sectional analysis of retinal layer
vasculature involvement. OMAG may be a useful modality to
characterize the retinal vasculopathy of Susac’s syndrome.
Commercial Relationships: Raghu C. Mudumbai, None; Qinqin
Zhang, None; Chieh-Li Chen, None; Yanping Huang, None;
Ruikang K. Wang, Carl Zeiss Meditec (F), Carl Zeiss Meditec (P)
Support: Research To Prevent Blindness
Constriction of retinal arterioles by 3 and 12 months after gastric
bypass surgery in type 2 diabetic patients with poor glycaemic
control
Troels Brynskov1, 2, Caroline Schmidt Laugesen1, Torben Lykke
Sørensen1, 2. 1Department of Ophthalmology, Roskilde Hospital,
Roskilde, Denmark; 2Faculty of Health Science, University of
Copenhagen, Copenhagen, Denmark.
Purpose: Gastric bypass surgery induces large metabolic changes
and rapidly normalizes blood glucose levels in type 2 diabetic
patients, but the consequences for the retinal vasculature are largely
unknown. We hypothesized that gastric bypass surgery could lead to
early changes in retinal vessel diameters – a marker of endothelial
function previously associated with the development of diabetic
retinopathy – in a cohort of patients with type 2 diabetes.
Methods: This was a prospective, observational clinical study. A
total of 53 eyes of 53 patients underwent thorough ophthalmologic
baseline examination 2 weeks before Roux-en-Y gastric bypass
surgery. All patients had type 2 diabetes. The examinations were
repeated at 3 and 12 months after the surgery. Retinal arterioles and
venules were semi-automatically measured using the “big-six”method on a fundus picture taken of the optic disk. We analyzed the
data using paired t-tests. Subgroup analysis was performed using a
linear regression model adjusting for sex, age, change in body mass
index, change in mean arterial pressure (MAP), baseline HbA1c, and
change in HbA1c.
Results: Higher HbA1c at baseline led to a decrease in the central
retinal artery equivalent (CRAE) of 2.2 mm per % HbA1c (±2.0,
p=0.04) at 3 months and of 3.0 mm per % HbA1c (±2.4, p=0.01) at 12
months. A fall in blood pressure compared to baseline was associated
with a decrease in both CRAE (p=0.006) and central retinal vein
equivalent (CRVE, p=0.01) at 3 months but did not reach significance
at 12 months (p=0.051 and p=0.12 respectively). All retinal vessel
diameter measures for the whole group were unchanged: at 3 months
after surgery CRAE had decreased by a mean of 1.3 mm (±1.9,
p=0.17) while CRVE had increased by a mean of 0.59 mm (±2.7,
p=0.66) and the Arterio-Venous Ratio (AVR) decreased by a mean of
0.0042 (±0.011, p=0.44). At 12 months CRAE decreased a mean of
0.95 mm (±2.2, p=0.39), CRVE decreased a mean of 2.2 mm (±2.7,
p=0.12) and AVR increased a mean of 0.0015 (±0.011, p=0.79).
Conclusions: Patients with poorer glycaemic control as measured by
HbA1c at baseline experienced an early and sustained constriction
of the retinal arterioles in the first year after gastric bypass surgery.
This is an encouraging finding, as other longitudinal studies have
identified arteriolar constriction as an early marker for improvement
in diabetic retinopathy.
Commercial Relationships: Troels Brynskov, None; Caroline
Schmidt Laugesen, None; Torben Lykke Sørensen, None
Support: Fight for Sight, Denmark
Clinical Trial: SJ-205
Assessment of Macular Circulation in Patients With Retinal
Vasculitis using OCT Angiography
Rasanamar Sandhu, Yali Jia, Liang Liu, Neal V. Palejwala, Eric B.
Suhler, Thomas S. Hwang, David Huang, Phoebe Lin. Casey Eye
Institute, Oregon Health Sciences University, Portland, OR.
Purpose: A feature of OCT angiography is its ability to provide
quantitative estimate of retinal blood flow by calculating vessel
density. This study examines macular blood flow using OCT
angiography in eyes with angiographically active retinal vasculitis
compared to normal eyes.
Methods: Adult patient with retinal vasculitis were imaged with
fluorescein angiography (FA) and OCT angiography with 70 kHz
OCT, using the split-spectrum amplitude decorrelation angiography
algorithm (SSADA). A 3 x 3 mm angiogram centered at the fovea
was obtained by projecting the flow signal internal to the RPE in
the en face orientation. Parafoveal vessel density was defined as
percentage of pixels with detectable flow signal in a 1mm-wide ring
surrounding the fovea (Fig. 1A). The choriocapillaris vessel density
was calculated as a percentage of pixels with detectable flow signal
within 10 microns external to the RPE in the 3x3mm area.
Results: 5 patients (7 eyes) with angiographically active retinal
vasculitis were included in the study. Their diagnoses included lupus
retinal vasculitis with choroiditis, Bechet’s disease, TINU with retinal
vasculits, sarcoidosis, and idiopathic retinal vasculitis. 11 normal
eyes were drawn from a prevously compiled database. The average
vessel density in normal eyes, in a 1mm wide parafoveal ring (Figure
1A), was 87.1% (95% CI 83.9-90.2). In eyes with retinal vasculitis,
the average parafoveal vessel density (see Figure 1B for example)
was significantly lower, at 79.8% (95% CI 76.3-83.4, p=0.006). We
also imaged choroidal blood flow in the patient with lupus vasculitis
and choroiditis. Her choriocapillaris vessel densities were 83.3% and
83.6% in the right and left eyes, respectively, compared to 96.3%
(95%CI 94.0%-98.5%) in normal eyes (n=7).
Conclusions: Patients with retinal vasculitis have significantly lower
parafoveal vessel density compared to normal eyes, as measured by
OCT angiography. Lower parafoveal vessel density was noted even
in patients who had only peripheral vasculitis on FA. This technique
shows promise as a possible biomarker for determining disease
activity, and gauging treatment response in patients with retinal
vasculitis.
Fig1A: OCT angiography of a normal eye. The circles demarcate the
1mm-wide parafoveal ring where vascular density was measured.
density was 6.5% and 1.5% coefficient of variation (CV) respectively.
In the vasculitis participants, the peripapillary retinal flow index was
0.085 ± 0.005 (mean ± SD), which was significantly less (P<0.05)
than that of the normal group (0.094 ± 0.006). The vessel density in
the vasculitis participants was 85.2% ± 1.3% (mean ± SD), which
was significantly less (P<0.01) than that of the normal group (90.9%
± 1.3%)
Conclusions: High quality OCT angiograms of peripapillary retina
could be obtained in both normal and retinal vasculitis participants.
Retinal vasculitis reduction in peripapillary perfusion could be
visualized as focal defects and quantified as flow index and vessel
density. OCT angiography could be useful in the clinical evaluation
of retinal vasculitis.
Fig1B: OCT angiography of a patient with lupus retinal vasculitis,
displaying lower parafoveal vascular density.
Commercial Relationships: Rasanamar Sandhu, None; Yali
Jia, Optovue, Inc. (F), Optovue, Inc. (P); Liang Liu, None; Neal
V. Palejwala, None; Eric B. Suhler, None; Thomas S. Hwang,
None; David Huang, Carl Zeiss Meditec, Inc (P), Optovue, Inc. (F),
Optovue, Inc. (I), Optovue, Inc. (P); Phoebe Lin, None
Support: T32EY023211, R01 EY023285, R01 EY024544, DP3
DK104397, RPB, CTSA grant (UL1TR000128)
Angiography of Peripapillary Retina in Retinal Vasculitis with 70
kHz Spectral OCT
Liang Liu1, 2, Yali Jia1, David Huang1, Phoebe Lin1, Alex D.
Pechauer1, Eric B. Suhler1. 1casey eye institute, Oregon Health &
Science University, Portland, OR; 2Ophthalmology, Peking Union
Medical College Hospital, Beijing, China.
Purpose: To compare peripapillary retinal perfusion between normal
and retinal vasculitis subjects using a commercially available optical
coherence tomography (OCT) system
Methods: Each study participant was imaged using a 3x3 mm
angiography scan by a high-speed (70 kHz) 840 nm spectrometerbased OCT system. The split-spectrum amplitude decorrelation
angiography (SSADA) algorithm was used to compute angiograms.
The peripapillary retinal flow index and vessel density were
calculated in the 700mm wide annulus extending outward from the
optic disc boundary. The flow index was defined as the average
decorrelation value on the retinal angiogram in the annulus region.
The peripapillary retinal vessel density was defined as the percentage
area occupied by vessels. Mann–Whitney test was used to compare
these perfusion variables between vasculitis patients and normal
subjects
Results: The study included 7 normal(7 eyes) and 4 retinal vasculitis
patients(5 eyes). In the normal eye, a dense microvascular network
around disc was visible on OCT angiography (Fig. B). This network
was visibly attenuated in the vasculitis eyes, and focal capillary
dropout was detected (Fig. E, F). In normal participants, the
population variability of peripapillary retinal flow index and vessel
Results from a normal eye(A, B) and a retinal vasculitis
eye(C,D,E,F). The 3x3mm peripapillary En face OCT angiogram(E)
of the region indicated by the red square in C and D, the microvasular
network was reduced compared with normal eye(B). The supranasal
OCT angiograms(F) of the region indicated by the yellow square in
D, the yellow arrow shows focal capillary dropout in the vasculitis
eye(F)
Commercial Relationships: Liang Liu, None; Yali Jia, Optovue,Inc
(F), Optovue,Inc (P); David Huang, Carl Zeiss Meditec,Inc (P),
Optovue,Inc (F), Optovue,Inc (I), Optovue,Inc (P); Phoebe Lin,
None; Alex D. Pechauer, None; Eric B. Suhler, None
Support: R01 EY023285, R01 EY024544, DP3 DK104397, RPB,
CTSA grant (UL1TR000128).
Optical Coherence Tomography Angiography of the Peripapillary
in Response to Hyperoxia
Alex D. Pechauer1, Yali Jia1, Liang Liu1, Simon S. Gao1, Chunhui
Jiang2, David Huang1. 1Ophthalmology, Oregon Health & Science
University, Portland, OR; 2Department of Ophthalmology, Eye and
ENT Hospital, Fudan University, Shanghai, China.
Purpose: Compare the peripapillary perfusion in healthy subjects
before and after hyperoxia using a commercially available optical
coherence tomography (OCT) system.
Methods: Participants were imaged after a 10 minute exposure
to normal and then hyperoxic breathing conditions. One eye of
each subject was scanned twice by a high-speed (70 kHz) 830 nm
wavelength spectrometer-based OCT system. The optic disc region
was scanned using a 3x3 mm volumetric scan. The split-spectrum
amplitude decorrelation angiography (SSADA) algorithm was used to
compute 3D angiograms. Horizontal and vertical-priority scans were
registered and merged to obtain one motion-corrected angiogram (Fig
1). The flow index (FI) was the average decorrelation value of the
peripapillary on the en face angiogram. The vessel density (VD) was
the percent area occupied by vessels in the peripapillary.
Results: Six healthy participants were scanned. The FI at baseline
was 0.108 ± 0.011 (mean ± SD), which was significantly more
(P = 0.001, T-test) than hyperoxia (0.099 ± 0.011). There was a
significant difference (P = 0.007, T-test) in VD between baseline
(95.9 ± 2.23) and hyperoxia (93.3 ± 3.43). Repeatability coefficient
of variation (CV) for baseline FI was 5.75% and for VD 1.67%. The
reproducibility CV for baseline FI and VD was found to be 11.1%
and 1.14%, respectively. Each participant had a large variation
in between-day autoregulatory response (Fig 2). The hyperoxia
induced average percent change relative to the baseline mean had a
reproducibility CV of 44.7% for FI and 75% for VD.
Conclusions: Using SSADA OCT we have shown that peripapillary
microvasculature blood flow can be measured under both normal and
hyperoxic conditions using a commercially available OCT system.
The decrease in peripapillary perfusion in response to an increase
in oxygen partial pressure provides further evidence of retinal
autoregulation. The autoregulatory response varied between days.
Fig.1. Angiograms at baseline (A) and hyperoxia (B). Image (B)
shows a 17% decrease in flow index and a 4% decrease in vessel
density.
Fig. 2. Vessel density (A) and flow index (B) average from each
subject. Percent change in vessel density (C) and flow index (D)
during hyperoxia at day one and two.
Commercial Relationships: Alex D. Pechauer, None; Yali Jia,
Optovue (F), Optovue (P); Liang Liu, None; Simon S. Gao, None;
Chunhui Jiang, None; David Huang, Carl Zeiss Meditec (P),
Optovue (F), Optovue (I), Optovue (P)
Support: R01 EY023285; R01 EY024544; DP3 DK104397; T32
EY23211; CTSA grant (UL1TR000128); RPB
Use of ICG-loaded erythrocytes for choroidal angiography in
human, pilot study
Giulia Caminiti1, Susanna Maria Carta1, Robert Flower3, Luigina
Rossi2, Mauro Magnani2, Maurizio Fossarello1, Enrico Peiretti1. 1Eye
Clinic, University Of Cagliari, Cagliari, Italy; 2University, Urbino,
Italy; 3Oftalmology, University school of Medicine, New York, USA
Minor Outlying Islands.
Purpose: Evaluation of the safety and efficacy of a new methodology
of retinal and choroidal dynamic angiography using human
erythrocytes preloaded with indocyanine green dye.
Methods: A group of 5 patients with different retinal diseases
(3 affected by diabetic retinopathy, 1 affected by central serous
chorioretinopathy and 1 with exudative age related maculopathy) and
a control group from our team of 3 healthy people with no history of
eye diseases underwent to an SLO dynamic ICG angiography, using
Heidelberg HRA Spectralis (Heidelberg Engineering).
After a withdrawn of 50 ml of autologous blood, red blood cells
(RBCs) were processed in a sterile manner using a specific medical
device: the Red Cell Loader (EryDel)® with the CE approval.
The procedure consisted in the dialysis of erythrocytes at a high
hematocrit (about 70-80%) against a hypotonic saline solution
(dialysis buffer) to allow the opening of membrane pores of the
RBCs; subsequently RBCs were incubated in the presence of the
substance to be encapsulated (ICG). The final step of our loading
phase was the resealing process by the restoration of physiological
isotonicity. The processing time was about 2 hours.
After the preparation of the RBC loaded with ICG, a bolus of
these cells was then reinjected in the antecubital vein of the same
withdrawn patient. Different concentrations (from 1 ml to 5 ml)
of our RBCs loaded were randomly injected before in the healthy
controls in order to explore the best quantity for a good angiography.
Before and after the angiography with RBCs loaded, the following
parameters were measured: blood pressure, hematocrit.
Results: High-speed ICG angiography showed the movement of
individual and clusters of ICG-loaded erythrocytes in the retinal
and the choroidal vessels. The signal was more visible in the early
frames than in the late frames. None of the patients involved in the
study showed any ophthalmic or general adverse events after the
intravenous injection of ICG-loaded erythrocytes.
Conclusions: This pilot trial suggests that the angiography with the
ICG-loaded RBCs is safe and well tollerated, meanwhile is very easy
to perform. The methodology should be improved on a higher sample
of patients in order to verify the rationality and any future clinical
applications.
Commercial Relationships: Giulia Caminiti, None; Susanna
Maria Carta, None; Robert Flower, None; Luigina Rossi, None;
Mauro Magnani, None; Maurizio Fossarello, None; Enrico
Peiretti, None
Imaging the morphology, rheology and flux of single red blood
cells in the living mouse eye without contrast agents
Jesse B. Schallek1, Andres Guevara-Torres2, 1, David R. Williams1, 2.
1
Center for Visual Science, University of Rochester, Rochester, NY;
2
The Institute of Optics, University of Rochester, Rochester, NY.
Purpose: Adaptive optics scanning light ophthalmoscopy (AOSLO)
has been used to measure blood velocity in the living retina by
tracking displacement of single blood cells. However, the complex
morphology of blood cells has not yet been characterized due to
insufficient cell boundary contrast. Here, we use differential imaging
to resolve the shape of single red blood cells (RBCs) in the retinal
circulation without using contrast agents.
Methods: Anesthetized C57BL/6J mice were imaged with an
AOSLO using near infrared light. The confocal pinhole in the
detection arm was replaced by a split-detector configuration (Scoles
et al. 2014 IOVS) where the left and right half of the imaging point
spread function was diverted into two, phase-locked photomultiplier
tubes (PMT). Differencing the PMT signals provided differential
contrast. 2-D point scanning at 25Hz was used to image slow moving
RBCs. 1-D point scanning across a vessel at ~31 kHz provided high
temporal resolution to image RBCs as they crossed the imaging
beam.
Results: We observed RBC deformation as cells 6.5 μm in diameter
squeezed through the ~4 μm vessel lumen in the smallest capillaries
(fig 1a). RBCs maintained a biconcave surface despite a high
deformation index (length/diameter) that ranged from 1.55-2.42,
similar to those reported in other tissues. 31 kHz 1-D scanning across
a vessel imaged a train of erythrocytes that could be counted (RBC
flux). Capillaries ranged from 40-161 cells/s (RBC volume of 1.9-7.7
picoliters/s). Capillaries showed robust modulations in RBC flux
that corresponded to the heart rate of the anesthetized mouse (~300
beats/minute) demonstrating that pulsatile flow is pervasive in the
smallest vessels (fig 1b). The leading and trailing edge of moving
RBCs displayed classic “parachute” and “slipper” morphologies (fig
2a) revealing the microscopic rheology of RBC interactions with the
vascular endothelium, plasma and glycocalyx. Capillaries showed
heterogeneity in RBC packing density despite having similar velocity
and size (fig 2bc).
Conclusions: This near infrared approach provides new
hemodynamic information in capillaries while mitigating phototoxic
exposure and obviating the need for blood contrast agents that may
alter hemodynamics. Future studies analyzing the shape of moving
RBCs have the potential to provide differential diagnosis in a variety
of systemic diseases without requiring a blood draw.
Commercial Relationships: Jesse B. Schallek, University of
Rochester (P); Andres Guevara-Torres, Canon Inc. (F), University
of Rochester (P); David R. Williams, Canon Inc. (F), Canon Inc. (R),
Polgenix Inc. (F), University of Rochester (P)
Support: Support: F32 EY023496, BRP EY014375. The Schmitt
Program on Integrative Brain Research Postdoctoral Fellowship, and
Canon Inc.
Conjunctival blood flow velocity in patients with retinal vasculitis
assessed with the retinal function imager
Nicole Stuebiger1, Aizhu Tao2, 3, Wen-Hsiang Lee2, Sandra Pineda2,
Hong Jiang2, Jianhua Wang2, Janet L. Davis2, Delia DeBuc2.
1
Department of Ophthalmology, Charite, University Medicine Berlin,
Berlin, Germany; 2Department of Ophthalmology, University of
Miami, Bascom Palmer Eye Institute, Miami, FL; 3Ophthalmology
and Optometry, School of Wenzhou Medical College, Wenzhou,
China.
Purpose: Because studies of the conjunctival microvasculature
have provided sensitive indicators of both systemic and CNS
vascular diseases we herein present the feasibility and applicability
in diagnostic imaging of conjunctival blood flow (BF) velocity in
patients with retinal vasculitis using a commercially available Retinal
Function Imager (RFI, Optical Imaging Ltd, Rehovot, Israel).
Methods: The RFI is a fundus camera-based device and was
developed primarily for imaging the retinal BF velocity. To test
the feasibility of using the RFI to assess the BF velocitiy of the
conjunctival microvasculature in retinal vasculitis patients, 12
patients (10 patients with Birdshot Chorioretinopathy; 2 patients
with retinal vasculitis of unknown origin) (n=21eyes; m:f=3:9, aged
52.4±12.2years) were recruited and compared with a healthy control
group (11 individuals; n=11eyes; m:f=3:8; aged 45.5±10.8years). The
temporal conjunctiva was imaged in each subject in one or both eyes
by RFI and high-resolution, non-invasive capillary perfusion maps
(nCPMs) were assessed in addition. The retinal BF of the right eyes
(ODs) was imaged for comparison.
Results: In the control group the conjunctival blood flow velocity in
ODs was 0.80±0.17mm/s and in the left eyes (OSs) 0.77±0.15mm/s.
Comparing these data with the conjunctival BF velocities of the
study group, we achieved significant differences. The vasculitis
patients disclosed conjunctival BF velocity in ODs of 0.71±0.07mm/s
(p<0.001) and a BF velocity in OSs of 0.74±0.06mm/s (p<0.05). In
the control group BF velocities of the retinal arteries and veins were
3.67 mm/s and 2.32 mm/s, in the study group 2.57 mm/s (p<0.00006)
and 1.56mm/s (p<0.00001), respectively. The microvasculature
anatomy revealed by the nCPMs appeared unevenly distributed,
and lower number of blood vessels along with lower degree of
complexity of their branching patterns were evident when compared
with a normal healthy eye.
Conclusions: With the RFI we could demonstrate for the first
time, that in patients with reduced retinal BF velocity due to retinal
vasculitis also the conjunctival BF is significantly impaired and
mirror the BF velocity changes of retinal microvasculature. Thus,
imaging the conjunctival vasculature with the RFI could offer an
easily to assess diagnostic tool in retinal vasculitis patients.
Commercial Relationships: Nicole Stuebiger, None; Aizhu Tao,
None; Wen-Hsiang Lee, None; Sandra Pineda, None; Hong Jiang,
None; Jianhua Wang, NIH (F), RBP (F); Janet L. Davis, None;
Delia DeBuc, Department of Defense (F), NIH (F), NIH Center Core
Grant (F), US 61/139,082 (P)
A pilot study of OCT angiography of iris melanomas
Alison Skalet1, Yan Li1, Chen D. Lu2, Yali Jia1, ByungKun Lee2,
Joachim Hornegger3, James G. Fujimoto2, David Huang1. 1Casey
Eye Institute, Oregon Health and Science University, Portland, OR;
2
Electrical Engineering & Computer Science, Massachusetts Insitute
of Technology, Cambridge, MA; 3Pattern Recognition Lab and
SAOT, University Erlangen Nuremberg, Erlangen, Germany.
Purpose: The purpose of this pilot observational clinical study was
to evaluate a new, noninvasive OCT angiography technique in the
imaging of iris melanomas.
Methods: The eyes of two patients who were diagnosed with iris
melanoma were evaluated using a swept-source, anterior segment
OCT system operating at 1050 nm wavelength and 100 kHz axial
scan repetition rate. Three-dimensional OCT angiography data
was acquired over 6 mm x 6 mm regions with scan depth of 5 mm
in tissue by using 3 repeated B-scans at 300 raster positions, each
B-scan consisting of 300 axial-scans. Horizontal and vertical raster
scans were acquired and software motion correction was applied to
reduce eye motion and combine the volumes. The split-spectrum
amplitude-decorrelation angiography (SSADA) algorithm was used
to detect flow and construct angiograms. En face OCT angiograms
were constructed by maximum flow projection.
Results: OCT angiography detected tortuous and disorganized
vascular patterns within pigmented iris melanomas in two eyes
(Figure 1). In normal areas of iris outside of the region of tumor
involvement, and in normal control eyes, vessels were oriented
radially. The tumor vasculature appeared denser than the vessels in
unaffected iris areas.
Conclusions: OCT angiography at 1050 nm can successfully image
vasculature within pigmented iris tumors and may be a less invasive
(no injection) alternative to conventional fluorescein angiography for
assessing tumors vascularity and monitoring response to treatment.
This is the first demonstration of OCT angiography in ocular tumors
and further studies are needed.
Commercial Relationships: Alison Skalet, None; Yan Li, Carl
Zeiss Meditec (P), Optovue (F), Optovue (P); Chen D. Lu, None;
Yali Jia, Optovue (F), Optovue (P); ByungKun Lee, None; Joachim
Hornegger, Carl Zeiss Meditec (P), Optovue (P); James G.
Fujimoto, Carl Zeiss Meditec (P), Optovue (F), Optovue (P); David
Huang, Carl Zeiss Meditec (P), Optovue (F), Optovue (I), Optovue
(P)
Support: Supported by R01EY018184, R01EY023285,
UL1TR000128, R01EY011289, a Lloyd Research Endowment
Faculty Grant and an unrestricted grant from Research to Prevent
Blindness.
414 Novel ophthalmic imaging
Wednesday, May 06, 2015 8:30 AM–10:15 AM
Contributing Section(s): Cornea, Retina, Visual Psychophysics/
Physiological Optics
Relationship between visual brain connectivity and duration of
blindness depends on onset of visual deprivation
Kevin C. Chan1, 2, Matthew C. Murphy1, 2, Christopher Fisher2,
Seong-Gi Kim1, 4, Joel S. Schuman2, 3, Amy C. Nau2. 1NeuroImaging
Laboratory, University of Pittsburgh, Pittsburgh, PA; 2UPMC
Eye Center, Eye and Ear Institute, Ophthalmology and Visual
Science Research Center, Department of Ophthalmology, School
of Medicine, University of Pittsburgh, Pittsburgh, PA; 3Department
of Bioengineering, Swanson School of Engineering, University
of Pittsburgh, Pittsburgh, PA; 4Center for Neuroscience Imaging
Research, Institute for Basic Science, Sungkyunkwan University,
Suwon, Korea (the Republic of).
Purpose: Visual deprivation is known to induce plasticity of
the visual system, which can be observed through alterations in
functional brain connectivity (FC) by functional MRI (fMRI).
How these FC changes accrue over time in congenital and acquired
blindness remains uncertain. This work aimed to model the effects of
prior visual experience on visual FC in blind subjects.
Methods: Seven congenitally blind and 11 age-matched acquired
blind subjects underwent 8 minutes of fMRI at rest using a 3 Tesla
scanner. A visual FC map was constructed for each subject by
computing the average correlation coefficient between the extrastriate
visual cortex and each voxel in the brain. We then examined the
effects of visual experience by fitting these FC maps with a linear
model with predictors including duration of blindness, a dummy
variable for congenital blindness, and the interaction of these two
effects. T score maps of these effects were thresholded at a familywise error corrected p<0.01.
Results: From the maps of voxel-wise statistical testing (Fig. 1),
more than 6 times as many voxels in the brain show a significant
group by duration interaction effect (row 4) compared to a simple
correlation with duration of blindness (row 2), indicating that
the relationship between visual FC and duration of blindness is
significantly different between congenital and acquired blindness.
In general, the sign of the correlation between FC and duration
of blindness in congenital subjects is opposite to that in acquired
subjects. Similar findings are observed when using striate cortex
instead of extrastriate cortex as the seed region of interest (data not
shown).
Conclusions: This work represents an early step toward
understanding plasticity in the visual system and how it depends upon
prior visual experience. These results suggest that alterations in FC
due to visual deprivation progress over time but in opposite directions
between congenital and acquired blindness. Our findings indicate that
longitudinal measures of FC and not only FC alone may be essential
for characterizing the state of the visual system.
Figure 1. Summary of voxel-wise statistical testing (FWE corrected
p<0.01). Row 1: T score for average visual FC map for all subjects.
Row 2: T score for significant relationship between FC and duration
of blindness. Row 3: T score for group-wise differences. Row 4: T
score for group by duration interaction.
Commercial Relationships: Kevin C. Chan, None; Matthew C.
Murphy, None; Christopher Fisher, None; Seong-Gi Kim, None;
Joel S. Schuman, Zeiss, Inc. (P); Amy C. Nau, None
Support: NIH Grant P30 EY008098 and T32-EY017271 (Bethesda,
MD); US Department of Defense DM090217; Alcon Research
Institute Young Investigator Grant (Basel, Switzerland); Eye and
Ear Foundation (Pittsburgh, PA); Research to Prevent Blindness
(New York, NY); Aging Institute Pilot Seed Grant, University of
Pittsburgh; Louis J. Fox Center for Vision Restoration of UPMC and
the University of Pittsburgh
In vivo MR Diffusion Weighted Imaging at 7 Tesla – feasibility
study in healthy subjects and patients with different ophthalmic
disease
Tobias Lindner1, Katharina Paul2, Andreas Graessl2, Jan Rieger3, Till
Huelnhagen2, Paul-Christian Krueger4, Anselm Juenemann5, Soenke
Langner4, Thoralf Niendorf2, 3, Oliver Stachs5. 1Preclinical Imaging
Research Group, University Medicine Rostock, Rostock, Germany;
2
Berlin Ultrahigh Field Facility, Max-Delbrueck Center for Molecular
Medicine, Berlin, Germany; 3MRI Tools, Berlin, Germany; 4Institute
for Diagnostic Radiology and Neuroradiology, University Medicine
Greifswald, Greifswald, Germany; 5Department of Ophthalmology,
University Medicine Rostock, Rostock, Germany.
Purpose: Diffusion-weighted imaging (DWI) of the eye and orbit
is an emerging MRI application for ophthalmic research, preclinical
imaging and diagnostic radiology. DWI probes and depicts selfdiffusion of water in tissue on a microscopic scale and holds
the promise to enhance diagnostic accuracy and provide further
physiological information for different ophthalmic disease. The
purpose of this study was to show the feasibility of in vivo DWI at 7
T with diffusion sensitized multi-shot RARE sequences.
Methods: In vivo MR imaging and DWI was performed at 7T in
healthy subjects (n=5), patients with uveal melanoma and/or retinal
detachment (n=5) and a patient with optic nerve glioma. A dedicated
six element transmit and receive coil array was employed at 7.0 T
and high resolution T1 weighted 3D FLASH imaging (TR=710.3ms,
TE=3.6ms, spatial resolution = (0.3x0.3x1.0)mm3 and T2-weighted
RARE imaging (TR=2940ms, TE=54ms, spatial resolution =
(0.25x0.28x1.0)mm3 were conducted for anatomical imaging. For
ADC mapping diffusion sensitized multi-shot RARE was used (7.0 T:
TR=3000ms, TE=133ms, spatial resolution = (0.4x0.4x3.0)mm3). For
diffusion sensitization b-values ranging from b=0 s/mm2 to b=400 s/
mm2 were employed. To reduce eye movement during imaging the
subjects followed an optoacoustic trigger protocol.
Results: The in vivo ADC map of patients with ocular mass
clearly delineates hypointense tumor from the surrounding modest
hypointense hemorrhage and the hyperintense vitreous body. These
ADC differences induced an ample contrast between the subretinal
hemorrhage and the tumor. ADC map of orbita in a patient with optic
nerve glioma showed ample ADC contrast versus the subarachnoid
space. The ADC map of the eye and optic nerve yielded a delineation
of the tumor which is superior to the contrast in the T2 weighted
images.
Conclusions: This work demonstrates the feasibility of in vivo
MR microscopy and diffusion sensitized ms-RARE of the eye and
optic nerve at 7.0T including T1w, T2w anatomical imaging along
with DWI and ADC mapping. This study showed that ms-RARE
based techniques offer immunity to B0 inhomogeneities and hence
are particularly suited for ophthalmic DWI. This is of clinical
relevance since single-shot echo planar imaging is prone to magnetic
susceptibility artifacts induced by the air filled nasal cavities and
frontal sinuses surrounding the eye.
Commercial Relationships: Tobias Lindner, None; Katharina
Paul, None; Andreas Graessl, None; Jan Rieger, MRI Tools (E);
Till Huelnhagen, None; Paul-Christian Krueger, None; Anselm
Juenemann, None; Soenke Langner, None; Thoralf Niendorf,
MRI Tools (S); Oliver Stachs, None
Fundus Changes in Highly Myopic Eyes with Different Shapes
Identified by High-resolution Three-dimensional Magnetic
Resonance Imaging
Xinxing Guo1, Ou Xiao1, Yanxian Chen1, Mingguang He1, 2. 1Division
of Preventive Ophthal, Zhongshan Ophthalmic Center, Guangzhou,
China; 2Centre for Eye Research Australia, University of Melbourne,
Melbourne, VIC, Australia.
Purpose: High myopia is often associated with elongation and
distortion of the globe, and at greater risk of developing pathologic
changes. Lack of topographic and morphologic assessments hinders
the understanding towards the natural history and mechanism in its
development. We evaluated the fundus changes in high myopia by
analyzing the eye shapes using three-dimensional (3D) magnetic
resonance imaging (MRI) in a case series study.
Methods: A total of 95 patients with high myopia (spherical power ≤
-6.00D) in both eyes underwent a series of ophthalmic examinations.
Eye shapes were categorized into spherical, ellipsoidal, cylindrical,
nasal-distorted, temporal-distorted and barrel-shaped according
to the inferior view from T2-weighted 3D MRI images (Achieva
3.0T, Philips Medical Systems, Best, the Netherlands); posterior
staphyloma was also identified. Fundus lesions were determined
using two 45° photos centered at macula and optic nerve by dilated
fundus photography (Canon Inc., Tokyo, Japan). The distributions of
various fundus characteristics in different eye shape categories were
analyzed using one-way ANOVA and Mann-Whitney Wilcoxon tests.
Results: The participants consisted 48.4% females, had a mean age
of 32.0±14.0 years old, spherical equivalence of -11.00±7.44D and
axial length (AL) of 28.18±1.73mm. The same ocular shapes between
the two eyes were observed in 72.6% of the patients. The most
predominant shape in the right eye was spherical (54.3%), followed
by nasal-distorted (16.0%) and cylindrical (14.9%). Barrel-shaped
eyes had the longest AL and most outstanding fundus changes.
Diffuse chorioretinal atrophy and larger peripapillary atrophy (PPA)
were more frequently observed in barrel-shaped (100.0%, 4.7±2.7
disc area, DA) and nasally distorted (73.3%, 2.5±2.5 DA) eyes;
While diffuse (77.3%) and patchy (22.7%) chorioretinal atrophy, as
well as fundus staphyloma (72.7%) were more common in eyes with
posterior staphyloma (n=22) determined by 3D MRI.
Conclusions: Barrel-shaped eyes present most significant visionthreatening conditions, while eyes with posterior staphyloma display
more severe chorioretinal atrophy. Our findings suggest different
topographic and morphologic patterns may be involved in the
development of high myopia.
Different eye shapes and corresponding fundus images
Commercial Relationships: Xinxing Guo, None; Ou Xiao, None;
Yanxian Chen, None; Mingguang He, None
Eye morphology quantitated by magnetic resonance imaging in
C57Bl/6 mice
Tomasz J. Choragiewicz1, Michal Fiedorowicz2, Marlena WelniakKaminska2, Jaroslaw Orzel2, 3, Piotr Bogorodzki2, 3, Robert Rejdak1,
2
, Pawel Grieb2. 1Department of General Ophthalmology, Medical
University of Lublin, Lublin, Poland; 2Mossakowski Medical
Research Centre, Polish Academy of Sciences, Warsaw, Poland;
3
Faculty of Electronics and Information Technology, Warsaw
University of Technology, Warsaw, Poland.
Purpose: Evaluating eye dimensions is crucial in studies of animal
models of eye diseases. However, performing these measurements
in alive mice is challenging. The aim of this study was to develop a
method for quantitating eye morphology using planar surface MRI
coil.
Methods: Aged (18 months) C57Bl/6 female mice (n=5) were
anaesthetized with isoflurane (4% in oxygen – induction, 1.52% - maintenance) and placed in 7T small animal-dedicated
magnetic resonance tomograph (BioSpec 70/30USR; Bruker
BioSpin, Ettlingen, Germany). Small planar surface receive coil
(internal diameter=10 mm, Bruker BioSpin, Ettlingen, Germany)
was placed over left eye of the imaged animal. High resolution
structural imaging with TURBORARE T2 sequence (TR=2700ms,
TEeff=30ms, RARE factor=8, NA=12, spatial resolution=0.062mm x
0.062mm, slice thickness=0.3mm, slices=7, scan time=16min) was
performed. Basic ocular dimensions and anterior chamber angle were
measured manually using OsiriX software (Pixmeo, SARL, Bernex,
Switzerland).
Results: Mean anterior chamber depth in C57Bl/6 mice was 0.464
mm (SD ±0.040) and anterior chamber angle (ACA) 24.98° (SD
±5.599), vitreous chamber depth (VCD) 0.465 mm (SD ±0.033),
axial length (AL) 3.586 mm (SD ±0.079) and horizontal length (HL)
3.394 mm (SD ±0.061) of the eyeball, lens thickness (LT) 2.146 mm
(SD ±0.059) and optic nerve diameter (ON) 0.250 mm (SD ±0.0520),
retinal thickness (RT) 0.206 mm (SD ±0.018).
Conclusions: High resolution magnetic resonance imaging of the
eye with use of small planar coil provide reproducible and consistent
measures of key dimensions of eyeball. These results are consistent
with results obtained with optical coherence tomography in C57Bl/6
mice [Chou et al. 2011, IOVS 52:3604-3612].
Commercial Relationships: Tomasz J. Choragiewicz, None;
Michal Fiedorowicz, None; Marlena Welniak-Kaminska, None;
Jaroslaw Orzel, None; Piotr Bogorodzki, None; Robert Rejdak,
None; Pawel Grieb, None
Support: The study was supported by Polish National Science
Centre grant No. 2012/07/D/NZ4/04199, Tomasz Choragiewicz was
supported by Allergan European Retina Panel II.
Through the Eyes of the Retinal Surgeon: Google Glass for
Recording Scleral Buckling Surgery
Michelle E. Wilson, Ehsan Rahimy, Sunir J. Garg. Ophthalmology,
Wills Eye Hospital, Philadelphia, PA.
Purpose: Google Glass (Google Inc., Mountain View, CA), with
its point-of-view (POV)-style recording, has the potential to
revolutionize the field of medical education, allowing trainees to see
the various steps of surgical procedures directly from viewpoint of
the operating surgeon. Unlike intraocular surgery, which benefits
from the video documenting capabilities of modern ophthalmic
surgical microscopes, scleral buckling surgery has traditionally been
difficult to record, limiting access to a valuable educational modality.
We report our experience using the Google Glass to record scleral
buckling surgery.
Methods: We recorded and subsequently reviewed both still images
and videos of 3 primary scleral buckling surgeries. We qualitatively
assessed the ability of the recorded material to document key steps in
the scleral buckling procedure, noting overall image quality as well
as factors limiting image quality.
Results: Google Glass was effective in recording the operative field
during each step of the scleral buckling procedure. Still images
(Figure 1) and video clips (Video, Supplemental) obtained provided
sufficient detail to demonstrate the key steps of the procedure.
Optimal image quality was limited by several factors including overor under- illumination and lack of magnification.
Conclusions: The field of medical education has benefited from
new technologies that allow trainees to visualize complex surgical
procedures from the viewpoint of the primary surgeon. The recent
introduction of Google Glass extends this capability to surgical
procedures that do not require an operating microscope and have
traditionally been difficult to record for the purposes of medical
education. Although limited by lighting and magnification, the
point of view recordings created using Google Glass during retinal
buckling surgery are able to document the key steps of the scleral
buckling procedure, providing a valuable tool for surgical teaching.
Scleral buckling surgery as recorded by Google Glass
Commercial Relationships: Michelle E. Wilson, None; Ehsan
Rahimy, None; Sunir J. Garg, None
Real-Time Swept-Source Microscope-Integrated versus HandHeld Spectral-Domain Optical Coherence Tomography during
Macular Hole Surgery
Michael I. Seider, Oscar Carrasco-Zevallos, Brenton Keller, Joseph
A. Izatt, Cynthia A. Toth, Christian Viehland. Duke University,
Durham, NC.
Purpose: To compare the clinical utility of a novel real-time
swept-source microscope-integrated optical coherence tomography
(SS-MIOCT) system to hand-held spectral-domain OCT (HH-OCT)
during macular hole repair.
Methods: All patients undergoing macular hole repair from 9/1/2014
– 10/31/2014 at the Duke Eye Center by a senior Vitreoretinal
Surgeon (CT) who gave consent were included. During macular
surgery, SS-MIOCT was used for all patients and allowed realtime imaging through the operating microscope. The system has
acquisition and processing rates of 2-10 volumes per second and
permitted the creation of volumetric renderings of surgical maneuvers
over time.
HH-OCT (Bioptigen Inc., Morrisville, NC, USA) was also performed
in the operating room at pauses in surgery immediately before and
at the completion of macular surgery. The images produced by both
systems were compared post-operatively.
Results:
Five female and 1 male patient with average age 70 years were
included, contributing three right and three left eyes in total.
SS-MIOCT was able to obtain high-quality, volumetric renderings
of surgical maneuvers in real-time without causing an interruption
in surgery. Particularly striking images documented instruments
brushing the retinal surface and forceps peeling internal limiting
membrane (ILM). SS-MIOCT scans allowed for the characterization
of anatomy-instrument relationships such as the distance of an
instrument from the retinal surface, the retinal deformation created
during scraping and the characteristics of elevated membranes
volumetrically and over time.
Hand-held OCT produced high-resolution two-dimensional scans
at pauses during surgery. These images allowed for the accurate
characterization of the morphology of the macular holes before and
after ILM peeling, and the confirmation of removal of epiretinal
membranes. For the cases of ILM autografting, HH-OCT created
useful images confirming the location of the autograft over the
macular hole.
Conclusions: SS-MIOCT appears to have significant advantages
over HH-OCT during macular hole surgery. SS-MIOCT was able to
capture volumetric scans of surgical maneuvers in real-time without
interrupting surgery whereas HH-OCT was useful for obtaining highquality static images of retinal anatomy during breaks in surgery.
Commercial Relationships: Michael I. Seider, None; Oscar
Carrasco-Zevallos, None; Brenton Keller, None; Joseph A. Izatt,
None; Cynthia A. Toth, None; Christian Viehland, None
Real-time 4D Stereoscopic Visualization of Human Ophthalmic
Surgery with Swept-Source Microscope Integrated Optical
Coherence Tomography
Oscar Carrasco-Zevallos1, Brenton Keller1, Christian Viehland1,
Liangbo Shen1, Gar Waterman1, Crystal Chukwurah1, Paul Hahn2,
Anthony N. Kuo2, Cynthia A. Toth2, 1, Joseph A. Izatt1, 2. 1Biomedical
Engineering, Duke University, Durham, NC; 2Opthalmology, Duke
University Medical Center, Durham, NC.
Purpose: Ophthalmic surgery is performed with a microscope that
provides limited depth perception. Surgeons often rely on indirect
cues for depth information. Current intraoperative Spectral-Domain
OCT systems are limited to cross-sectional real-time imaging.
Ophthalmic surgery is performed in a 3D surgical field; therefore,
a real-time 3D micron-scale imaging modality could be useful for
surgical feedback and guidance. We report on the development of
4D (volumetric imaging + time) Microscope Integrated OCT (4D
MIOCT) for real-time volumetric visualization of human ophthalmic
surgery.
Methods: The MIOCT sample arm enabled concurrent OCT and
operating microscope imaging. The system employed a custom
swept-source OCT engine operating at 1060 nm. GPU-based custom
software enabled real-time acquisition, processing, and rendering
of volumetric images at 100k A-lines/second. Volumetric frame
rates varied between 2-10 Hz. A custom microscope integrated
stereoscopic heads-up display (HUD) allowed for visualization
of MIOCT volumes through the surgical binoculars. 4D MIOCT
imaging was performed in 25 human surgeries (7 anterior segment,
18 vitreoretinal). Vitreoretinal cases imaged included macular
hole, retinal detachment, and epi-retinal membrane (ERM)
procedures. Anterior segment cases imaged included cataract, deep
anterior lamellar keratoplasty, and Descemet stripping endothelial
keratoplasty procedures.
Results: Figure 1 shows 4D MIOCT imaging of an ERM peel with
surgical forceps during a macular hole surgical case. The time stamps
for each volume in the time series are in seconds. B-scans are shown
below each corresponding volume. The location of the B-scans is
denoted by the white rectangle on the volume view. Surgical camera
frames are shown as well with a yellow rectangle denoting the
MIOCT field of view. Figure 2 illustrates 4D MIOCT imaging of a
cataract case. A-B depict volumes and B-scans acquired at different
stages of cataract surgery. (A) shows the intact cataract. (B) shows
cataract fragmentation with a phacoemulsification needle (red arrow).
(C) shows the anterior segment after cataract removal. (D) shows
intraocular lens (green arrow) insertion.
Conclusions: Real-time, volumetric, micron-scale visualization of
human ophthalmic surgery was performed with 4D MIOCT.
4D MIOCT imaging of an ERM peel.
4D MIOCT imaging of cataract surgery.
Commercial Relationships: Oscar Carrasco-Zevallos, None;
Brenton Keller, None; Christian Viehland, None; Liangbo Shen,
None; Gar Waterman, None; Crystal Chukwurah, None; Paul
Hahn, None; Anthony N. Kuo, Bioptigen (P); Cynthia A. Toth,
Alcon (P), Bioptigen (F), Duke University (P), Genetech (F); Joseph
A. Izatt, Bioptigen (I), Bioptigen (P), Bioptigen (S)
Support: NIH EY023039
Clinical Trial: NCT01588041
Impact of Microscope Integrated OCT on ophthalmology
resident performance of anterior segment maneuvers in model
eyes
Bozho -. Todorich1, Christine Shieh1, Philip DeSouza1, Oscar
Carrasco-Zevallos2, David Cunefare1, Joseph A. Izatt2, Sina Farsiu1,
Prithvi Mruthyunjaya1, Anthony N. Kuo1, Cynthia A. Toth1, 2.
1
Ophthalmology, Duke University, Durham, NC; 2Bioengineering,
Duke University, Durham, NC.
Purpose: The integration of swept-source optical coherence
tomography (SS OCT) in the operating microscope enables realtime, tissue-level imaging to aid in ophthalmic microsurgery. In
this prospective randomized controlled study, we evaluated the
impact of SS microscope-integrated OCT technology (MI OCT) on
ophthalmology residents’ performance in anterior segment surgery.
Methods: Ophthalmology residents (N=14) were recruited, stratified
by year of training, and randomized to perform the following four
anterior segment maneuvers on porcine eyes without (Group A)
or with (Group B) direct MI OCT guidance: 50% and 90% depth
corneal suture passes, repair of vertical linear corneal laceration via
suturing at 90% thickness, and creation of a tri-planar clear corneal
incision. Both groups of residents then repeated the maneuvers
without MI OCT guidance. Lastly, the non MI-OCT controls (Group
A) repeated all maneuvers under direct MI OCT guidance.
Volumetric OCT scans acquired at the end of each maneuver were
manually segmented to compute point of maximal depth of each
corneal pass and tri-planar corneal incision profile. Differences
between Group A and B were compared using one-way ANOVA
(Figure 1A). Subjective feedback through a survey was also obtained
from each resident after they had completed all maneuvers (Figure
1B).
Results: Figure 1 shows the results from Group A (no MIOCT
guidance) and Group B (MIOCT guidance). Residents operating
with direct MI OCT feedback demonstrated enhanced performance
in depth-based anterior segment maneuvers compared to the control
group (p<0.0001). Residents trained with MI OCT continued to
outperform the controls when operating without direct MI OCT
feedback (p<0.0001) (Fig 1A). Surgical performance of residents
who were trained and tested without MI OCT improved to equivalent
performance when given MI OCT feedback. Resident surgeons
rated subjective experience of using MI OCT very favorably and on
average are “more likely” to use it in their future practice (Fig 1B).
Conclusions: Real-time MI OCT guidance improved depth based
surgical maneuvers. MI OCT may be a useful adjunct in surgical
education of ophthalmology residents.
Purpose: To assess the retinal macroarchitectural alterations that
occur during membrane peeling for vitreomacular interface (VMI)
disorders utilizing intraoperative OCT (iOCT)
Methods: The PIONEER study is a prospective study examining
iOCT in ophthalmic surgery. Eyes undergoing membrane peeling
were evaluated with video/iOCT correlation. Clinical variables
assessed included membrane peeling technique (e.g., diamond-dusted
membrane scraper [DDMS], forceps). iOCT images were reviewed
for macroarchitectural changes (e.g., inner and full-thickness retinal
changes). A microscope mounted portable SD-OCT probe (Bioptigen,
Research Triangle Park, NC) was used to obtain preincision and
post-peel iOCT images. Postoperative OCT images were analyzed at
1 week and 1 month.
Results: Of the total 163 eyes, post-peel retinal changes were noted
by iOCT in 45 eyes (28%). Inner retinal elevations were noted in 38
eyes (23%), and 8 eyes (5%) had full-thickness retinal elevations.
Based on video/iOCT correlation, 95% of the retinal changes
correlated with instrument utilization. Direct peel initiation resulted
in 47% of the changes and indirect pulling during completion of
peel resulted in 48% of the changes. The retinal elevations occurred
in 33% of the eyes when using forceps-only and in 23% of the eyes
when using both forceps and a DDMS (p= 0.16, chi square). At
1-week post-op, 19% of the retinal changes persisted as focal inner
retinal thickening or thinning. No full-thickness retinal changes
persisted. At 1-month post-op, 60% of inner retinal thickening had
resolved, while all focal retinal thinning persisted.
Conclusions: Significant subclinical macroarchitectural changes
occur during membrane peeling for VMI conditions that are
directly related to instrument manipulations. iOCT technology is
able to identify these changes. There was a trend towards increased
alterations with forceps-only technique compared to using a DDMS
for peel initiation. Inner retinal changes persisted in many eyes at
1 month postoperatively. Further research is needed on the visual
effects of these changes and how to minimize these changes during
surgery.
Figure 1. Surgical performance (A) and subjective feedback (B) of
ophthalmology residents with and without MI OCT feedback.
Commercial Relationships: Bozho -. Todorich, None; Christine
Shieh, None; Philip DeSouza, None; Oscar Carrasco-Zevallos,
None; David Cunefare, None; Joseph A. Izatt, Bioptigen
(I), Bioptigen (P), Bioptigen (S); Sina Farsiu, None; Prithvi
Mruthyunjaya, None; Anthony N. Kuo, Bioptigen (P); Cynthia A.
Toth, Alcon (P), Bioptigen (F), Genentech (F)
Support: R01-EY02039 (Toth/Izatt) and Research to Prevent
Blindness grant (DeSouza)
Macroarchitectural Intraoperative Retinal Alterations Following
Membrane Peeling Visualized with Intraoperative OCT in the
PIONEER Study
Jaehong Han1, 2, Sunil K. Srivastava2, Peter K. Kaiser2, Rishi P.
Singh2, Justis P. Ehlers2. 1Case Western Reserve University School
of Medicine, Cleveland Heights, OH; 2Ophthalmic Imaging Center,
Cole Eye Institute of Cleveland Clinic, Cleveland, OH.
A: preincision iOCT image; B: post-peel iOCT image; C: iOCTgenerated fundus map; D: correlated surgery video
A, B: preoperative OCT-generated fundus map and OCT image; C,
D: postoperative OCT-generated fundus map and OCT image
Commercial Relationships: Jaehong Han, None; Sunil K.
Srivastava, Allergan (R), Baucsh and Lomb (C), Baucsh and Lomb
(R), Bioptigen (P), Carl Zeiss Meditec (C), Leica (C), Synergetics
(P); Peter K. Kaiser, Alcon (C), Bausch and Lomb (C), Bayer (C),
Carl Zeiss Meditec (C), Genentech (C), Novartis (C), Ophthotech
(C), Oraya (C), Regeneron (C), Topcon (C); Rishi P. Singh, Carl
Zeiss Meditec (C); Justis P. Ehlers, Alcon (C), Bioptigen (C),
Bioptigen (F), Bioptigen (P), Genentech (F), Leica (C), Leica (F),
Synergetics (P), Thrombogenics (C), Thrombogenics (F), Zeiss (C),
Zeiss (F)
Support: NIH/NEI K23-EY022947-01A1 (JPE); Ohio Department
of Development TECH-13-059 (JPE); Thrombogenics research grant:
THROM1403JE
Evaluation of a Surgical Microscope Interfaced SD-OCT system
for Anterior Segment Surgery
Marco Ruggeri1, Florence Cabot1, Sonia H. Yoo1, Jean-Marie A.
Parel1, 2. 1Ophthalmology, Bascom Palmer Eye Inst, Univ of Miami,
Miami, FL; 2Vision Cooperative Research Centre, Brien Holden
Vision Institute, UNSW, Sydney, NSW, Australia.
Purpose: To evaluate the use of an intraoperative SD-OCT system
that attaches to an ophthalmic operating microscope (OPMI) for ondemand OCT guided anterior segment (AS) surgery.
Methods: We used a commercial prototype (Bioptigen, Inc.)
that consists of a compact OCT delivery console connected to a
transportable cart that houses an SD-OCT engine, a computer for
image acquisition and processing and a monitor for image display.
The system provides near 9mm axial resolution over an imaging
depth of 15.4 mm in air. The scanning console can attach to several
commercial OPMI models and includes a proprietary microscope
objective with a focal length of 175 mm, a two axes OCT scanning
mechanism and the optics to combine the OCT scanning beam with
the illumination and viewing paths of the host OPMI. The delivery
console is setup by removing the original microscope objective lens
from the host OPMI and by securing the scanning head enclosure to
the dovetail connection of the host OPMI. Prior to surgery, the OCT
delivery unit was attached to the OPMI and the cart was deployed
near the bedside so that intraoperative OCT images of the surgical
site could be displayed in real-time to the surgeon using the computer
monitor.
Results: Under an IRB approved protocol, 4 patients who underwent
AS surgical procedures including cataract and DSAEK surgeries
were imaged intraoperatively. Surgeries were uneventful. The system
enabled displaying the images to the surgeon without interrupting
the surgical maneuvers or blocking the OPMI view. The distance
of 167 mm between the microscope objective and the patient eye
provided enough room for the surgeon to operate comfortably. The
compact surgical OCT interface did not interfere with the sterile field
and the standard operations performed by the surgical staff. Preoperatively, the system enabled to confirm that the AS did not show
any abnormality. Throughout DSAEK surgeries, the long imaging
depth and the volumetric capability of the SD-OCT system enabled
to verify optimal adhesion of the entire graft to the host cornea in 3D.
Throughout cataract surgery the system enabled to verify the integrity
of the capsule and the final position of the intraocular lens.
Conclusions: The intraoperative SD-OCT system can be easily
and safely used to assist the surgeon for AS surgeries. This tool
has potential to improve the outcome of DSAEK and help with
challenging cases of cataract surgery.
Commercial Relationships: Marco Ruggeri, Bioptigen Inc. (F);
Florence Cabot, None; Sonia H. Yoo, None; Jean-Marie A. Parel,
Bioptigen Inc. (F)
Support: Bioptigen Inc; NIH P30EY14801 (Center Grant); Florida
Lions Eye Bank; Drs KR Olsen and ME Hildebrandt, Research to
Prevent Blindness; The Henri and Flore Lesieur Foundation (JMP).
Real-time dynamic depth tracking for arbitrarily long range
OCT imaging and surgical instrument tracking using a Fourier
domain optical delay line
Mohamed El-Haddad, Yuankai Tao. Ophthalmic Research, Cleveland
Clinic Foundation, Shaker Heights, OH.
Purpose: Intraoperative optical coherence tomography (iOCT)
allows visualization of tissue microstructure, and provides real-time
feedback for clinical decision making during ophthalmic surgery.
However, the potential for iOCT-guided surgical maneuvers has
been limited both by imaging speed, which can only display single
cross-sectional images at video-rates, and imaging range, which is
on the order of 2 mm for SDOCT and 8 mm for SSOCT. Last year,
we presented novel lateral tracking technology that automatically
centered the iOCT scan field on the tip of ophthalmic surgical
instruments. Here, we demonstrate a digitally controlled optical
delay that allows real-time depth tracking. Integration of automated
lateral and depth tracking with iOCT will allow for dynamic threedimensional field-of-view imaging of ophthalmic surgical dynamics.
Methods: A reference arm based on a Fourier domain optical delay
line (FDODL), was designed to provide a digitally controlled optical
delay for iOCT. A free-space optical delay on a motorized stage was
added to the FDODL path to further extend the available tracking
range. A control algorithm handles the actuation hand-off between the
stage and the FDODL to allow for fast scanning over a short range,
and slower scanning over a long range (Fig. 1). The FDODL allowed
inter-line depth tracking over >16 mm depth range with a 100 ms
small-angle step response (Fig. 1, δz) while the motorized stage
allowed for inter-frame scanning over a 100 mm range at speeds up
to 100 mm/s (Fig. 1, Δz).
Results: As a demonstration, we imaged a metal surface placed at
~62 deg. to the horizontal (Fig. 1). Without tracking, small part of the
surface was visible (Fig. 1b, blue). The FDODL was then scanned by
a saw-tooth function, effectively compensating for the tilt (Fig. 1b,
red). The A-Scans in the latter represent different depths, in linear
proportion to the applied voltage (3.7 mm/volt), which was corrected
for in post-processing to reflect the true image over the entire range
of the FDODL.
Conclusions: Integrating automated lateral and depth tracking with
iOCT will allow real-time imaging of surgical dynamics. Improved
understanding of tissue-instrument interactions during conventional
ophthalmic surgical maneuvers may elucidate mechanisms of tissue
repair, be predictive of clinical outcomes, and lead to novel iOCTguided surgical techniques.
from all ocular structures: epiretinal membranes, internal limiting
membrane, outer retinal layers.
Our purpose is showing, directly through videos, the easy use of
iOCT and its add-on, like checking the closure of macular holes
during surgery ( high myopic macular hole closure by external
indentation), the complete removal of ERM and ILM even without
using dyes twice.
Methods: ZEISS RESCAN Lumera 700 (Carl Zeiss Meditec, Dublin,
Calif., USA) - with integrated intra-operative OCT technology and
Callisto eye software (Carl Zeiss Meditec, Dublin, Calif., USA)
can be used during all surgical procedures performed on anterior
segment, glaucoma, and retina. The wavelength of spectral domain
iOCT is 840 nm, the scanning speed is 27000 A-scans per second
with an A-scan depth of 2.0 mm, an axial resolution of 5.5 μm in
tissue. Scan length is adjustable from 3 to 16 mm, and scan rotation
is adjustable over 360°. Image acquisition is possible by recording
movies or taking pictures.
Results: The new machinery, iOCT, improves the safety and the
surgeon’ self confidence during every surgical procedure.
Starting new era that opens other surgical scenarios: surgeons can
work with anatomical guide. iOCT shows anatomical reperies
of the eye directly in surgical field overlaying the main image.
It is controlled by the foot switch, allowing the surgeon to move
indipendently from other members of the surgical staff thus
enhancing the theatre safety.
Conclusions: In conclusion, the importance of our video is related
to the use of a new technique that allows the use of OCT during
surgery and in selected cases of diagnosis as well. Further studies are
certainly necessary for a better comprehension of uses and limitations
of iOCT, but our videos for the first time demonstrate possible new
applications.
iOCT showing epiretinal membrane interrruption and its retraction
after sharp needle scraping.
Commercial Relationships: Michele Coppola, None; Federico
Solignani, None
Commercial Relationships: Mohamed El-Haddad, None; Yuankai
Tao, None
Utility of the intraoperative OCT during posterior segment
surgery
Michele Coppola, Federico Solignani. Ophthalmology, Azienda
Ospedaliera di Desio e Vimercate (MB), Milano, Italy.
Purpose: Intraoperative OCT is a new surgical tool helping surgeons,
improving intraoperative safety and allowing a real-time feedback
Intraoperative Optical Coherence Tomography (iOCT) In
Newborn And Infant Children With Anterior Segment Anomalies
Sebastian E. Siebelmann, Philipp Steven, Manuel Hermann, Thomas
Dietlein, Claus Cursiefen. Department of Ophthalmology, University
of Cologne, Cologne, Germany.
Purpose: Examination of newborn or infant children with anterior
segment anomalies is often challenging due to limited compliance, so
that an examination under general anesthesia is preferred. However,
intraoperative imaging tools are so far limited. Therefore we
evaluated intraoperative Optical Coherence Tomography (iOCT) as
a tool for intraoperative imaging, diagnostics and surgery monitoring
in newborn and infants with anterior segment anomalies and corneal
opacities.
Methods: Retrospective case series of four newborn and infants
with anterior segment anomalies (2 children with congenital corneal
opacities, 1 child with congenital cataract and one with iridocorneal
synechiae), who underwent examinations under general anesthesia,
using a commercially available 840nm OCT, mounted to an operating
microscope (iOCT; OptoMedical Technologies, Luebeck, Germany).
Mean age was 6 years (range: 6 month – 14 years).
Results: Anterior segments could be completely visualized in the
intraoperative set up using iOCT under general anesthesia including
cornea, anterior chamber angle, iris and lens despite presence of
corneal opacities. iOCT was a helpful tool to detect corneal, iridal
and lens-anomalies in these newborn and infants, who were not
accessible to normal OCT-diagnostics without general anesthesia.
Moreover iOCT enabled intraoperative decision making to perform
anterior segment surgery (e.g. phototherapeutic keratectomy or
lamellar keratectomy) in patients with corneal opacities. Furthermore,
depth of tissue ablation could precisely be monitored during anterior
lamellar keratectomy.
Conclusions: Intraoperative Optical Coherence Tomography (iOCT)
closes the gap of imaging diagnostics observed in newborns and
infants, who have to be examined under general anesthesia and are
not accessible to “routine” advanced imaging techniques so far.
Furthermore iOCT proved to be a helpful tool for online monitoring
of anterior segment surgery in newborn and infant children.
Commercial Relationships: Sebastian E. Siebelmann, None;
Philipp Steven, None; Manuel Hermann, None; Thomas Dietlein,
None; Claus Cursiefen, None
Structural and functional retinal imaging with MHz Line-field
parallel swept source imaging (LPSI)
Daniel Fechtig1, 2, Tilman Schmoll3, Cedric Blatter1, Rainer A.
Leitgeb1, 2. 1Center for Med. Phys. and Biom. Eng., Medical
University Vienna, Vienna, Austria; 2Center for Med. Phys. and
Biom. Eng., Christian Doppler Society, Vienna, Austria; 33Carl Zeiss
Meditec, Inc., Dublin, CA.
Purpose: MHz OCT allows mitigating undesired influence of motion
artifacts during retinal assessment, but comes in state-of-the-art
point scanning OCT at the price of increased system complexity.
By changing the paradigm from scanning to parallel OCT for in
vivo retinal imaging the three-dimensional (3D) acquisition time
is reduced without a trade-off between speed, sensitivity and
technological requirements.
Methods: Line field parallel interferometric imaging (LPSI) is
utilizing a commercially available swept source, a single-axis
galvo-scanner and a line scan camera for recording 3D data with
up to 1MHz A-scan rate. Besides line-focus illumination and
parallel detection, we mitigate the necessity for high-speed sensor
and laser technology by holographic full-range imaging, which
allows for increasing the imaging speed by low sampling of the
optical spectrum. High B-scan rates up to 1kHz further allow
for implementation of lable-free optical angiography in 3D by
calculating the inter B-scan speckle variance.
Results: We achieve a detection sensitivity of 93.5 (96.5) dB at
an equivalent A-scan rate of 1 (0.6) MHz. Fig. 1(a) was obtained
by stitching 7 slightly overlapping non-averaged tomograms. The
sensitivity and resolution are high enough to visualize the external
limiting membrane (ELM), and to contrast the various layers of the
inner retina. Despite the loss of confocality in one lateral dimension
and the shorter center wavelength of 840nm, the signal intensity
from the choroid is remarkably strong maintaining structural details.
In Fig. 1(b) we feature a enface projection of micro-angiographic
information. One single 3D stack corresponds to a lateral FOV of
approx. 5-6° and is acquired in just 2.5 s.
Conclusions: Our results demonstrate for the first time competitive
imaging sensitivity, resolution and speed with a parallel OCT
modality. LPSI is in fact currently the fastest OCT device applied to
retinal imaging and operating at a central wavelength window around
800 nm with a detection sensitivity of higher than 93.5 dB.
Fig. 1(a) and 1(b)
Commercial Relationships: Daniel Fechtig, None; Tilman
Schmoll, None; Cedric Blatter, None; Rainer A. Leitgeb, None
Support: Christian Doppler Society
Investigation of the microscopic retina with MHz AO-OCT
Omer P. Kocaoglu, Tim L. Turner, Zhuolin Liu, Donald T. Miller.
School of Optometry, Indiana University, Bloomington, IN.
Purpose: Optical coherence tomography (OCT) has undergone
considerable technological advances over the last two decades, yet
image acquisition speed remains a major limiting factor for its use
in retinal imaging. Increased speed enables imaging of larger fields
of view at finer spatial and temporal sampling, while also reducing
the effects of eye motion. Reduction of eye motion is particularly
attractive with adaptive optics OCT (AO-OCT) because the cellularlevel lateral resolution afforded by AO makes the system more
susceptible to eye motion. In this study we employ a research-grade
AO-OCT system that acquires images 25 times faster than clinical
OCT and use it to investigate the microscopic retina in the living
human eye.
Methods: The Indiana MHz AO-OCT system is based on spectral
domain technology. It employs a high-speed 1×4 optical switch in
its novel, quad-spectrometer detection channel to achieve 1 million
A-lines/s acquisition speed. The design also makes efficient use of the
light available for detection. A superluminescent diode (λc=790nm,
Δλ=42nm) illuminates the retina providing 5.3 μm axial resolution.
The sample channel contains the AO system that dynamically
corrects for ocular aberrations across a 6.7 mm pupil to provide
diffraction-limited lateral resolution (1.7 μm confocal) and improved
signal-to-noise ratio of retinal images. Volumes videos were acquired
with the MHz AO-OCT system of all major retinal layers from
the retinal pigment epithelium layer to retinal nerve fiber layer at
6 superior to the fovea. Volumes were 1.1 ×1.3 or 0.5 ×0.7 in size
(containing 320 B-scans × 400 A-lines) and acquired at real time rates
of 7.8 Hz and in 30 volume sequences (3.84 s).
Results: MHz AO-OCT volumes were successfully acquired on three
normal subjects. Retinal images were typically of ~30 dB dynamic
range, without averaging, and revealed substantially reduced image
blur and distortion due to eye motion compared to earlier generation
AO-OCT systems developed at Indiana. The 30 dB dynamic range is
comparable to that of clinical OCT and permitted visualization of all
major retinal layers. The 7.8 Hz volume rate also captured the flow
dynamics of retinal capillaries at different depths in the inner retina.
Conclusions: MHz AO-OCT provides sufficient 3D resolution and
sensitivity to image at the microscopic level any major layer of the
retina.
display used for fixation. The monitor displayed videos and targets to
aid fixation during imaging.The swept-source OCT system employed
a 1060 nm frequency-swept laser (Axsun Tech; Billerica, MA) and a
Mach-Zender interferometer. The interferometric signal was detected
with a dual-balanced receiver (Thorlabs, Inc.; Newton, NJ) and
digitized at 800 MS/s (Alazar Tech Inc; QC, Canada). Custom GPUbased software enabled real-time volumetric imaging at 100,000
A-line/second.
Results: The axial resolution across the 6 mm imaging depth range
was measured to be 8.4um and the lateral resolution was measured
to be 11-12 um. The peak sensitivity was 99.4 dB with a 3.5 mm
-6dB falloff. To prove the feasibility of the system, consented adult
subjects were imaged. A representative volumetric and an averaged
B-scan image are shown in Fig 2. The subject was situated on a
chinrest 36 cm away from the system. The LCD monitor facilitated
fixation during imaging.
Conclusions: A novel long working distance OCT system along with
the fixation system were designed, built and tested on adult subjects.
The future implication is to image the retina of young children.
Representative MHz AO-OCT volumetric image displayed on a
linear-scale. Scale bars: 25μm.
Commercial Relationships: Omer P. Kocaoglu, None; Tim L.
Turner, None; Zhuolin Liu, None; Donald T. Miller, #7,364,296
(P)
Support: NEI grants R01-EY018339 and P30- EY019008
Long Working Distance Swept Source Optical Coherence
Tomography for Pediatric Imaging
Ruobing Qian1, Oscar Carrasco-Zevallos1, Lejla Vajzovic2, Cynthia
A. Toth2, 1, Joseph A. Izatt1, 2. 1Biomedical Engineering, Duke
University, Durham, NC; 2Ophthalmology, Duke University Medical
Center, Durham, NC.
Purpose: Optical coherence tomography (OCT) has become the
standard of care for diagnosis of many retinal pathologies. However,
current commercial OCT systems require cooperative patients to
maintain fixation for several seconds in a chinrest. Handheld OCT
systems have also been demonstrated for successful imaging of
supine patients, as well as pre-term infants and neonates up to ~1 year
old. However, no technology yet exists for OCT in young children
due to their lack of attention and cooperation, as well as inherent fear
of large objects close to their face. Therefore, we designed and built a
novel OCT system with a very long working distance (distance of the
last optical component of the system to the subject’s eye) to facilitate
imaging of young children.
Methods: A novel scanning configuration OCT system was designed
to achieve a working distance of 36 cm (Fig. 1) to situate young
children at a comfortable distance away during imaging. A 2f
scanning configuration, instead of the conventional 4f scheme, was
implemented to reduce the footprint and weight of the sample arm.
To optimize optical performance at the retinal plane, the 2f system
employed two custom-designed lenses (Zemax, Inc; Redmon, WA).
A dichroic mirror after the objective enabled co-alignment of a LCD
Commercial Relationships: Ruobing Qian, None; Oscar
Carrasco-Zevallos, None; Lejla Vajzovic, None; Cynthia A. Toth,
Alcon (P), Bioptigen (F), Duke University (P), Genetech (F); Joseph
A. Izatt, Bioptigen (I), Bioptigen (P), Bioptigen (S)
Support: The Hartwell Foundation
Birefringence analysis of retinal nerve fiber bundles in the human
eye
Christoph K. Hitzenberger1, Mitsuro Sugita1, 2, Michael Pircher1,
Bernhard Baumann1, Philipp K. Roberts3, Stephan Holzer3, Tomoyuki
Makihira2, Nobuhiro Tomatsu2, Makoto Sato2, Clemens Vass3. 1Center
F Med Physics & Biomed Eng, Medical University of Vienna,
Vienna, Austria; 2Canon Inc., Tokyo, Japan; 3Ophthalmology and
Optometry, Medical University of Vienna, Vienna, Austria.
Purpose: To analyze birefringence, retardation, and thickness of the
retinal nerve fiber layer (RNFL) along traces of retinal nerve fiber
bundles in the human eye in vivo.
Methods: A polarization sensitive (PS) OCT system with an
integrated retinal tracker was used to record 3D data sets in healthy
human eyes in vivo. En face maps of RNFL thickness, retardation,
axis orientation, and birefringence were generated from the PSOCT data sets. Based on the axis orientation and thickness maps,
traces of nerve fiber bundles were extracted. A new type of graphic
data representation was developed: a 2D false color map that plots
birefringence, retardation, and thickness along the nerve fiber bundle
(y-direction) as a function of azimuthal angle (x-direction) around the
optic nerve head (ONH). Based on these plots, a quantitative analysis
of the variations of parameters along the nerve fiber bundle is made.
Results: We analyzed the variations of retardation, thickness, and
birefringence along superior and inferior nerve fiber bundles in 4
healthy eyes. If measured from the ONH outward, retardation and
thickness showed a pronounced decrease along fiber traces near the
ONH of > 8°/mm and > 50 mm/mm, respectively, while birefringence
was rather constant, with a change of < 0.05°/mm/mm.
Conclusions: In healthy eyes, RNFL birefringence is rather constant
along the nerve fiber bundle length. Deviations from this pattern
might be indicative of diseases like glaucoma.
Wavefront sensorless adaptive optics OCT with real time axial
tracking for imaging human retina
Yifan Jian1, Kevin S. Wong1, Michelle Cua1, Stefano Bonora2, Robert
J. Zawadzki3, 4, Marinko V. Sarunic1. 1Engineering Science, Simon
Fraser University, Burnaby, BC, Canada; 2CNR-Institute of Photonics
and Nanotechnology, Padova, Italy; 3Department of Cell Biology
and Human Anatomy, University of California Davis, Davis, CA;
4
Department of Ophthalmology & Vision Science, University of
California Davis, Sacramento, CA.
Purpose: We describe our wavefront sensorless adaptive optics
optical coherence tomography (WSAO-OCT) system for imaging
the human photoreceptor mosaic in vivo. Real time segmentation of
the retinal OCT B-scan permitted axial tracking and extraction of en
face images. We validated our system performance by imaging the
retina at several eccentricities, and demonstrated the improvement in
photoreceptor visibility with WSAO aberration correction.
Methods: A human WSAO-OCT system was constructed using
lenses (instead of mirrors), delivering a 5.5mm (diameter) at the
subject’s pupil. The OCT engine acquired A-scans at a rate of 200
kHz, and volumes of 1024x200x80 voxels were acquired in ~0.1
s. A custom written GPU program tracked the layer of interest
initially selected by the operator using a simple retinal segmentation
algorithm, and then extracted the corresponding en face projection
image. The modal WSAO algorithm optimized Zernike radial
orders 2 to 4 in order to maintain a balance between optimization
time and effective aberration correction. For each Zernike mode,
the optimization was performed by acquiring an OCT volume
and extracting an en face image for each of 5 different coefficient
values. The coefficient that produced the brightest en face image
was selected as the optimized value. The entire optimization process
required 6~12 seconds, depending on the amount of the aberrations in
the subject’s eye.
Results: We present en face images acquired at three different retinal
eccentricities before and after WSAO optimization in Figure 1. In
the unoptimized images, the cones are mostly indistinguishable from
speckle pattern. After optimization, the image contrast increased, and
the cone mosaic can be resolved.
Conclusions: We demonstrated a lens-based approach for WSAOOCT that is capable of resolving the cone mosaic in the human eye
at small angles of eccentricity with non-mydriatic pupils even with a
small-stroke DM. The reduced complexity of the lens-based WSAO
design can facilitate a robust and compact imaging system that is
highly suitable for clinical applications in ophthalmology.
Nerve fiber analysis of healthy human retina from PS-OCT data set.
Left: unfolded retardation map: retardation (color) is plotted along
nerve fiber bundle length (y-axis) as a function of origin on azimuthal
circle around ONH (x-axis). Right: unfolded birefringence map.
Commercial Relationships: Christoph K. Hitzenberger, Canon
Inc. (F), Canon Inc. (P); Mitsuro Sugita, Canon Inc. (E), Canon Inc.
(P); Michael Pircher, Canon Inc. (F), Canon Inc. (P); Bernhard
Baumann, None; Philipp K. Roberts, Canon Inc. (F); Stephan
Holzer, None; Tomoyuki Makihira, Canon Inc. (E); Nobuhiro
Tomatsu, Canon Inc. (E), Canon Inc. (P); Makoto Sato, Canon Inc.
(E), Canon Inc. (P); Clemens Vass, Canon Inc. (P)
Figure 1: En face images of the human photoreceptor layer acquired
at several angles of eccentricity from the fovea.
Commercial Relationships: Yifan Jian, (); Kevin S. Wong, None;
Michelle Cua, None; Stefano Bonora, None; Robert J. Zawadzki,
None; Marinko V. Sarunic, None
Support: CIHR, NSERC, Michael Smith Foundation for Health
Research, and Simon Fraser University Vice President of Research.
NEI (R01 EY024239), UC Davis RISE Grant, NSF I/UCRC CBSS
Grant. NVIDIA hardware donation program.
Cellular morphometry of the outer retina using adaptive optics
OCT
Ravi S. Jonnal, Justin Migacz, Iwona Gorczynska, Robert J.
Zawadzki, John S. Werner. Ophthalmology, UC Davis, Sacramento,
CA.
Purpose: Diseases and disorders which impact the outer retina, such
as AMD, diabetic retinopathy, macular hole, and retinal detachment
are routinely evaluated using optical coherence tomography (OCT).
As such, correct interpretation of clinical OCT images is of great
importance. Interpretation of these images, though, is clouded
by optical properties of the eye, the complexity of light-tissue
interactions, and image post-processing. We have designed and built
a custom OCT system equipped with adaptive optics that provides
greatly improved 3D resolution and permits precise instrument
calibration and characterization. We investigated the OCT outer
retinal band 3, thought to originate from either the cone outer
segment tips or zone of interdigitation between cone outer segments
(OS) and apical processes of the RPE.
Methods: Five subjects were imaged between 1.5 and 4.25°. Cones
were segmented and aligned by their band 3 peaks. The resulting
aligned B-scan was averaged and the thickness of band 3 was
measured. Next, after bulk motion correction, the phase of the band
3 reflection was extracted and variance of this phase, which is a
measure of the surface’s roughness, was computed.
Results: After subtracting measured axial blur, average thickness
of band 3, measured in single cones, was 2.1 μm. Typical variance
of bulk-motion-corrected phase measurements was 0.09 rad, which
corresponds to a surface roughness of 4 nm RMS.
Conclusions: Cellular measurements of band 3 suggest strongly that
the reflection comes from a thin, optically smooth origin, such as a
reflective surface, and not from an axially extended zone of scattering
material. This surface must be located distal to the cone OS lumen
and proximal to the RPE body. The most likely origin is the distal
plasma membrane of the cone outer segment.
AO-OCT B-scan taken at 1.5° temporal to the fovea. In (a), band 3
reflections from single cones are marked. Axial shifts of reflections
are evident. In (b), the cones are shown without axial alignment. In
(c), the cones are aligned by band 3.
Visualization of Multiple Retinal Capillary Beds using Offset
Pinhole Adaptive Optics Scanning Light Ophthalmoscopy
Richard B. Rosen1, 2, Nadim Choudury1, 2, Nikhil Menon1, 2, Alexander
Pinhas1, 2, Rishard Weitz1, Joseph Carroll3, Alfredo Dubra3,
Toco Chui1, 2. 1New York Eye and Ear Infirmary, New York, NY;
2
Ophthalmology, Ican School of Medicine at Mount Sinai, New York,
NY; 3Ophthalmology, Medical College of Wisconsin, Milwaukee, WI.
Purpose: To image retinal capillary beds at different retinal layers in
healthy and diseased retinas using an offset pinhole adaptive optics
scanning light ophthalmoscope (AOSLO).
Methods: Retinal capillary plexuses at different depths were imaged
at various retinal locations in 2 healthy controls and 3 patients (PDR,
HRVO, BRVO) using an offset pinhole AOSLO with an imaging
wavelength centered at 790nm. Imaging locations on the healthy
controls were located at 5° temporal, nasal, and superior to the fovea.
Region of interests with vasculopathic structures were pre-identified
on fundus pictures on the patients. Image sequences were acquired
using 1° or 1.5° field of view at a frame rate of 15Hz. After sinusoidal
distortion and eye motion were removed, averaged images and
motion contrast perfusion maps were generated at different retinal
depths.
Results: Multiple retinal capillary beds were clearly visualized
in both controls and patients using offset pinhole AOSLO. Three
distinct capillary beds were resolved at all retinal locations imaged
in the controls. From the 3 patients with retinopathies, vasculopathic
structures such as microaneurysms and tortuous blood vessels
were observed at different retinal depths. Fig 1 demonstrates a
fusiform microaneurysm located in the outer retina in a patient with
HRVO. This technique enabled precise depth location of normal
and pathologic structures with high lateral resolution that is often a
limiting factor when attempting similar stratification using optical
cohernece tomography
Conclusions: Offset pinhole AOSLO provides noninvasive and direct
visualization of multiple retinal capillary beds at different depths,
as well as the ability to isolate retinal vasculopathic structures at
different retinal capillary beds. This imaging technique provides a
better understanding of normal and pathological retinal vasculature
development.
(a) Bulk-corrected phase of band 3 of the cones in Fig. 1. Average
longitudinal reflectance profiles of the (b) unaligned cones and (c)
the band 3-aligned cones. Unaligned band 3 FWHM is 10.8 μm and
aligned FWHM is 4.1 μm, which includes the 2.8 μm (measured)
axial PSF. This suggests that the bulk of band 3’s thickness in
commercial OCT images is due to axial displacements of the
reflections.
Commercial Relationships: Ravi S. Jonnal, US Patent No.
7,364,296; unlicensed; (P); Justin Migacz, None; Iwona
Gorczynska, None; Robert J. Zawadzki, None; John S. Werner,
None
Support: NEI Grant R01 EY024239; unrestricted grant from
Research to Prevent Blindness, Incorporated
Fig 1 demonstrates a fusiform microaneurysm located in the outer
retina capillary layer in a patient with HRVO. The thru focus
technique is able to distinguish the precise level of the lesion
Commercial Relationships: Richard B. Rosen, Advanced Cellulat
Technologies (C), Allergan (C), Carl Zeiss Meditech (C), Clarity (C),
OD-OS (C), Opticology (I), Optovue (C); Nadim Choudury, None;
Nikhil Menon, None; Alexander Pinhas, None; Rishard Weitz,
None; Joseph Carroll, ImagineEyes (C); Alfredo Dubra, University
of Rochester (P); Toco Chui, None
Support: Marrus Foundation, Wise Foundation
Non-invasive adaptive optics imaging of the ground squirrel
retina
Benjamin S. Sajdak2, Christopher S. Langlo2, Yusufu N. Sulai1, Robert
F. Cooper3, Dana K. Merriman5, Joseph Carroll2, 1, Alfredo Dubra1,
4 1
. Opthalmology, Medical College of Wisconsin- Eye Institute,
Milwaukee, WI; 2Cell Biology, Neurobiology, & Anatomy, Medical
College of Wisconsin, Milwaukee, WI; 3Biomedical Engineering,
Marquette University, Milwaukee, WI; 4Biophysics, Medical College
of Wisconsin, Milwaukee, WI; 5Biology, University of Wisconsin
Oshkosh, Oshkosh, WI.
Purpose: Ground squirrels are an important model for studying
visual processing, retinal circuitry, and cone photoreceptor function.
Using a custom adaptive optics scanning light ophthalmoscope
(AOSLO), we sought to noninvasively image the microscopic
structure of the ground squirrel retina.
Methods: Four 13-lined ground squirrels (Ictidomys
tridecemlineatus) were anesthetized with isoflurane (5% induction,
2-3% maintenance) and placed on a heated stage. Pupils were
dilated with one drop of 2.5% phenylephrine and one drop of 1%
tropicamide. Saline drops were applied as needed to maintain
corneal hydration. Reflectance, confocal and split-detection AOSLO
imaging through a 4 mm diameter pupil was performed with near
infrared light (790 nm). The pixel size in microns was calculated
using a Ronchi ruling placed in the back focal plane of a 19 mm focal
length model eye, and then scaled according to the axial length of
each animal, measured with an ultrasound A-scan (OTI-Scan 1000).
Photoreceptor density and Voronoi geometry were calculated at
known retinal locations using cell coordinates obtained by a semiautomated cell counting algorithm.
Results: High-resolution images of the photoreceptor mosaic, nerve
fiber layer, and retinal vasculature (Fig.1) were obtained with a high
success rate (100%, n = 4). No lens opacities were observed during
any of the imaging sessions, which lasted up to 1 hour. Photoreceptor
density values are consistent with previously published histological
data of a congener species (Long and Fisher, J Comp Neurol., 1983;
Kryger et al., Vis Neurosci., 1998), ranging from 25,600–78,400
cells/mm2. The mosaic displayed a fairly triangular packing geometry,
with 42-56% of the cells having six-sided Voronoi domains around
the horizontal streak, and 59-69% in more peripheral locations.
Conclusions: The photoreceptor mosaic, retinal capillaries and
microscopic detail on the nerve fiber layer inner surface of the ground
squirrel were visualized with AOSLO. The image quality and success
rate appears to be better than current AOSLO efforts on mice and
rats. Thus, when combined with the non-invasive high resolution
imaging afforded through AOSLO, the ground squirrel could serve as
a useful model to aid drug discovery and testing through longitudinal
imaging on the cellular scale.
Ground squirrel cone photoreceptors using confocal (A) and split
detector (B) AOSLO. Nerve fiber layer (C) and retinal capilaries (D).
Scale bars 50mm
Commercial Relationships: Benjamin S. Sajdak, None;
Christopher S. Langlo, None; Yusufu N. Sulai, None; Robert F.
Cooper, None; Dana K. Merriman, None; Joseph Carroll, None;
Alfredo Dubra, None
Support: P30 EY001931, T32-GM080202
Spectral domain optical coherence tomography analysis
of the retina in a rat model: a comparison of transgenic
immunodeficient retinal degenerate rat (SD-Foxn1
Tg(S334ter)3Lav) and rats with normal retina
Alexander De Guzman1, 2, Bibo Khatib2, Robert Lin1, 2, Bryce
T. McLelland2, Anuradha Mathur2, Robert B. Aramant2, Brian
Cummings1, 2, Magdalene J. Seiler1, 2. 1Physical Medicine &
Rehabilitation, University of California, Irvine, Irvine, CA; 2Stem
Cell Research Center, University of California, Irvine, Irvine, CA.
Purpose: Retinal degeneration (RD) affects millions of people
worldwide, and in vivo monitoring of the progression of this disorder
is needed to aid the development of treatments for it, such as retinal
sheet transplants. The goal of this study is twofold: to show that
retinal degeneration worsens with age, and to demonstrate that
Spectral Domain Optical Coherence Tomography (SD-OCT) can
track the progression of this condition in a new immunodeficient RD
rat model.
Methods: Retinal cross-section SD-OCT scans (Bioptigen Envisu
R2200 Spectral Domain Ophthalmic Imaging System) were obtained
from male and female rats (n=40) aged between P17-P77, using a
group of transgenic SD-Foxn1 Tg(S334ter)3Lav rats (=RD rats) as a
model for retinal degeneration, and a normal retinal control group of
rats expressing human placental alkaline phosphatase (hPAP). Data
analysis was done using the Bioptigen post-scan analysis software
InVivoVue Diver Release 2.0 (Bioptigen, Research Triangle Park,
NC). Changes in the thickness of the total retina (TR), the outer retina
(OR), and the inner nuclear layer (INL) were determined with respect
to age. Using SigmaPlot 11.0 (Systat Software, Inc., San Jose, CA;
Fisher-corrected multiple comparison tests), layer thicknesses were
compared with age group and with rat strain.
Results: P17 was the earliest age in which retinas could be imaged.
At this age, there was already a noticeable difference in total retinal
(TR) and outer retinal (OR) thickness between hPAP rats and RD
rats. In normal rats, TR was decreasing with eye growth between
the ages of P17 and P77 by 10.4% (from 260μm to 233μm) whereas
TR of RD rats decreased by 32.6% (from 133μm to 95μm). OR
was decreasing with eye growth between the ages of P17-P77 by
6.5% (from 141μm to 132μm) whereas OR of RD rats decreased
by 28.8% (from 26μm to 20μm). The INL was decreasing with eye
growth between the ages of P17-P77 by 32.6% (from 38μm to 25μm)
whereas the INL of RD rats decreased by 38.8% (from 34μm to
21μm).
Conclusions: This study has demonstrated that SD-OCT is a
viable method of detecting and analyzing retinal degeneration; and
consequently, the status of the retina and any occurring changes can
be monitored and analyzed over time in vivo. SD-OCT may be useful
in the evaluation of treatments for retinal degeneration.
Commercial Relationships: Alexander De Guzman, None;
Bibo Khatib, None; Robert Lin, None; Bryce T. McLelland,
None; Anuradha Mathur, None; Robert B. Aramant, Ocular
Transplantation LLC (E), Ocular Transplantation LLC (P); Brian
Cummings, None; Magdalene J. Seiler, Ocular Transplantation
LLC (C), Ocular Transplantation LLC (P)
Support: CIRM Grant TR4-06648; Summer Undergraduate Student
Research Program, UC Irvine
Google Cardboard anterior and posterior segment imaging: a
valuable tool for limited-resource settings
Simon Ghofrani, Mahsa Rezaei, Aaron Wang, Allen O. Eghrari,
Christopher J. Brady. Wilmer Eye Institute, Johns Hopkins,
Baltimore, MD.
Purpose: In limited-resource settings such as developing countries,
significant obstacles to ocular imaging include cost and local
availability of required instruments. Smartphone photography
leverages the ubiquity of mobile handsets to image the posterior pole,
but typically requires the use of both hands to hold the device and
a lens, which generally must be purchased separately at significant
expense. Here, we describe the use of modified, open-source Google
Cardboard technology to image the anterior and posterior segment
using a head-mounted device composed of materials available
globally and at minimal cost.
Methods: A single Google Cardboard device, composed of cardboard
and acrylic lenses, was modified with a strap to mount to the
viewer’s head. Of the two lenses, each of which has a focal point of
4.5cm (22.2D) the right one was removed from the device and the
examiner’s smartphone placed within the device to be centered in
front of the left eye. A white LED light and coin cell battery were
affixed to the Cardboard to maximize smartphone battery life, and
installed superior to the mobile phone camera in order to provide
maneuverability comparable to an indirect ophthalmoscope.
Results: This device, plans for which are open source and
freely available online, allows the viewer to conduct indirect
ophthalmoscopy through a smartphone embedded in the headset,
while recording the examination from the perspective of the
examiner. The examination is facilitated with a 20D lens, but can
also be demonstrated using the detached right 22.2D acrylic lens
that is associated with the product, thereby allowing the examiner to
perform indirect ophthalmoscopy without costly equipment. Video
and clinical photographs of the technique are demonstrated.
Conclusions: Using only a smartphone, cardboard and acrylic,
products available in most regions of the world, indirect
ophthalmoscopy can be conducted through a headset and images
recorded for clinical or research use.
Still photo of indirect opthalmoscopy from video acquired through
Google Cardboard. Video to be demonstrated at time of presentation
reveals ease of use.
Commercial Relationships: Simon Ghofrani, None; Mahsa
Rezaei, None; Aaron Wang, None; Allen O. Eghrari, None;
Christopher J. Brady, None
A novel approach to ophthalmic photography using a portable
and versatile camera device
R Joel Welch, Quan Dong Nguyen. Ophthalmology, UNMC, Omaha,
NE.
Purpose: Smartphone fundus photography has grown in popularity
over recent years. The ability to use a portable device, such as a
smartphone, to capture high-resolution and clinically relevant fundus
photographs has had a significant impact on patient care and interphysician communication. While the cameras of modern smartphones
are growing increasingly robust, limitations do exist. To date, the
GoPro camera (GoPro Inc., San Mateo, CA, USA), considered as one
of the most powerful and versatile portable camera devices, has not
been used in ophthalmic photography. The purpose of this study is
to determine whether the GoPro camera can be used for ophthalmic
fundus photography.
Methods: A GoPro HERO4 Silver camera (GoPro Inc., San Mateo,
CA, USA ) and a 20D lens (Volk Optical Inc., Mentor, OH, USA)
were used for this study. Additionally, a battery-powered five
millimeter white LED (NTE Electronics, Inc., Bloomfield, NJ, USA)
was used as the coxial light source and attached to the camera using
Transpore medical tape (3 M, St. Paul, MN, USA). A simple and
inexpensive model of the human eye was constructed to aide the
author in learning image capture technique before approaching a
human volunteer. Once the technique was mastered using the model,
the right eye of one human volunteer was dilated in order to safely
capture human fundus images. GoPro software was then used for
image processing.
Results: The GoPro camera was able to capture images with both the
simple eye model and the healthy human eye. Images from the study
included healthy optic nerve, retinal vessel, and macular anatomy
(Figure).
Conclusions: The GoPro camera is capable of capturing fundus
images of the human eye; however, the clinical utility of the images
still needs to be demonstrated. The quality of the images rivals that
of smartphone ophthalmic photography and head-to-head testing
could be an area of future exploration. Additionally, while there was
success with the GoPro/LED combination for coaxial camera and
light source, improvements in the construct could potentially yield
higher image quality. High-quality ophthalmic images obtained
with relatively inexpensive and portable devices, such as the GoPro
camera, will likely continue to have a growing impact on the field of
ophthalmology.
Figure 1. Fundus photo of the right eye using GoPro camera device.
Commercial Relationships: R Joel Welch, None; Quan Dong
Nguyen, None
Slitlamp-mounted smartphone adapters using reverse
engineering with 3D scanning and 3D printing
Charline Boente1, Ethan Tu2, Rony Sayegh1. 1Ophthalmology,
University Hospitals Case Medical Center, Cleveland, OH;
2
Biomedical Engineering, Case Western Reserve University,
Cleveland, OH.
Purpose: Slitlamp smartphone adapters have become widely
available, however, only a limited number of phone models are
supported. We describe a process using 3D scanning and 3D printing
to build customizable adapters that fit virtually any smartphone to any
slitlamp.
Methods: With the selected smartphone, we obtained precise
dimensions of the phone using 3D scanning (NextEngine 3D Scanner,
Santa Monica, CA), which outputs a stereolithography (STL) file
that can be manipulated in a variety of design software programs.
We used AutoCAD software (AutoDesk, Inc. San Rafael, CA) to
design a custom adapter from the dimensions obtained from the
scanned phone using reverse engineering principles. A prototype of
the custom adapter was then produced using 3D printing (Stratasys
Fortus 400mc, Eden Prairie, MN). Each customized model was tested
on our clinics’ slitlamps and verified for fit and ease of obtaining
photographic images of the anterior segment.
Results: Slitlamp-mounted adapters for a variety of iPhone (Apple,
Cupertino, CA) models as well as the Samsung Galaxy S4 (Samsung
Electronics Co. Ltd., Ridgefield Park, NJ) were built using this
method. Precise measurements were obtainable using 3D scanning
which were used to build accurate and custom adapters using
AutoCAD followed by 3D printing.
Conclusions: 3D scanning provides easy data acquisition of the
phone dimensions that improves the accuracy of the 3D printed
adapters and allows a wider range of phone models to be used for
slitlamp photography.
Commercial Relationships: Charline Boente, None; Ethan Tu,
None; Rony Sayegh, None
Smartphone fundus photography, in vivo retinal fluorescent
photography and fluorescein angiography in mouse eyes
Cynthia X. Qian1, Eiichi Hasegawa1, Luis Haddock2, David M.
Wu1, 3, Shizuo Mukai1. 1Ophthalmology, Massachusetts Eye and Ear
Infirmary, Boston, MA; 2Bascom Palmer Eye Institute, Palm Beach
Gardens, FL; 3Genetics, Harvard Medical School, Boston, MA.
Purpose: To describe a technique of fundus photography, in vivo
fluorescent retinal photography and fluorescein angiography in mouse
eyes using a smartphone
Methods: Wild-type mice with normal fundi and with retinal
detachment and retinas expressing Green Fluorescent Protein
(GFP) or Discosoma Red Fluorescent Protein (DsRed) secondary
to adeno-associated virus (AAV) infection or electroporation. After
pharmacologic mydriasis, the mice were gently held in hand without
anesthesia. Using a modification of our techniques previously
described for human and rabbit eyes (Haddock et al. J Ophthalmol
2013, Article ID 518479) fundus photography was carried out using
an iPhone 5s® or iPhone 6® and a 78D lens (as compared to 20D
and 28D lenses for human and rabbit eyes respectively). For in vivo
fluorescent photography of retinas transduced with GFP or DsRed,
an excitation filter (but no barrier) for fluorescein angiography from
an old fundus camera was placed in front of the LED light source
but not the camera lens of the iPhone®. For fluorescein angiography,
an excitation filter was place in front of the LED and a barrier filter
was placed in front of the camera lens, and images were taken after
intraperitoneal injection of fluorescein.
Results: Using this system we were able to consistently take
high-quality video clips and fundus photographs in the eyes of
awake mice. The use of an excitation filter allowed for fluorescent
photography of the patches of subretinal GFP and DsRed transgene
expression. Using a combination of excitation and barrier filters we
were able to perform fluorescein angiography in wild-type mice.
Conclusions: We demonstrated that with a modification of a
previously described system for human and rabbit eyes, fundus
photography, in vivo retinal fluorescent photography and fluorescein
angiography can be performed using the iPhone®. This technique is
relatively inexpensive, readily available, and very portable.
Fluorescein angiography imaging highlighting normal vasculature in
a mouse retina
Comparison of Smartphone Ophthalmoscopy with Slit-lamp
Biomicroscopy for Undilated Glaucoma Screening
Andrea Russo1, Francesco Morescalchi1, Luisa Delcassi1, Ciro
Costagliola2, Francesco Semeraro1. 1Univ degli Studi di Brescia Italy, Brescia, Italy; 2Eye Clinic, University of Molise, Campobasso,
Italy.
Purpose: Smartphone ophthalmoscopy is a recent and promising
technique to screen the community. The purpose of this study is to
evaluate the agreement of smartphone ophthalmoscope and slitlamp biomicroscopy to discriminate eyes with glaucoma from those
without.
Methods: One hundred glaucoma patients and 100 healthy control
subjects were included. All of them underwent visual field test and
undilated smartphone ophthalmoscopy with D-Eye device (Si14
S.p.A., Padova, Italy) followed by dilated optic nerve head (ONH)
slit-lamp examination. The following parameters were evaluated:
cup/disc ratio; neuroretinal rim pallor (4-step scale); peripapillary
chorioretinal atrophy; and optic disc hemorrhages.
Results: All included eyes completed the study. Cup/disc ratio exact
agreement between the two methods was observed in 186 (93%)
of 200 eyes (simple κ = 0.91; 95% CI 0.87–0.95) and agreement
within one step was observed in 200 eyes (100%). Rim pallor exact
agreement was observed in 179 eyes (89.5%; simple κ = 0.85; 95%
CI 0.79–0.91) and agreement within one step was observed in 197
eyes (98.5%). An almost perfect agreement was observed for both
peripapillary chorioretinal atrophy and optic disc hemorrhages.
Conclusions: Undilated smartphone ophthalmoscopy showed a very
good agreement with dilated biomicroscopy for ONH assessment.
Smartphone ophthalmoscopy can accurately detect ONH parameters
and might be used as a screening tool for glaucoma.
Fluorescent photography highlighting patch of DsRed staining in a
rd1 mouse retina
Commercial Relationships: Cynthia X. Qian, None; Eiichi
Hasegawa, None; Luis Haddock, None; David M. Wu, None;
Shizuo Mukai, None
Representative retinal images of optic nerve head taken with D-Eye.
(A) Cup/disc ratio evaluated as 0.6 in a 76-year-old male patient
with nuclear cataract. (B) Healthy optic nerve head in a 23-year-old
female. (C) Depiction of the D-Eye prototype magnetically attached
to the smartphone.
Commercial Relationships: Andrea Russo, None; Francesco
Morescalchi, None; Luisa Delcassi, None; Ciro Costagliola, None;
Francesco Semeraro, None
Smart phone retinal photography: Uses and experiences with the
novel retinal imaging technique
Luis Leon, Jonathan I. Huz, Arkadiy Yadgarov, Marco A. Zarbin,
Neelakshi Bhagat. Ophthalmology, Rutgers New Jersey Medical
School, South Orange, NJ.
Purpose: Retinal imaging is an essential component in the
practice of ophthalmology; its use is limited to the availability
of necessary imaging equipment. We reviewed fundus photos
taken using a recently described technique of fundus photography
using a smartphone, and its utility in a large hospital with a busy
ophthalmology training program. This study describes and evaluates
the use of this technique in a setting when fundus cameras are not
available.
Methods: We performed a retrospective review of all fundus
photographs of retinal pathology taken using smart phone cameras
in the inpatient or emergency room settings. Fundus images were
captured with an iPhone 4, 5 or 6 and a 20D lens. Using the coaxial
light source from the phone and its included video software,the phone
functions as an indirect ophthalmoscope, high-definition videos of
the fundus can be recorded for subsequent still image extraction. The
pictures were evaluated by ophthalmologists to asses its quality and
clinical utility.
Results: We obtained fundus photos of patients with retinal
pathology incluiding retinal detachment, pre-retinal membranes,
proliferative diabetic retinopathy with pre-retinal and vitreous
hemorrhage, macular pucker, commotio and retinal hemorrhages
of shaken baby syndrome . Retina specialists the residents
easily identified the pathology that was photographed using this
technique. Pictures were of very good quality even when taken by
ophthalmology residents that were inexperienced with the technique.
The resolution and details observed in the photos was sufficient to
recognize important retinal details and to establish a diagnosis. The
pictures were also noted to be an excellent resource to compare
disease progression and changes of retinal anatomy in subsequent
examinations.
Conclusions: In our review, we noted that the technique to obtain
retinal pictures with smartphones is an excellent resource, especially
in the emergency room and inpatient setting. With this technique eye
doctors can assess and follow up retinal pathology effectively, high
quality fundus photos can be obtained by relatively inexperienced
personnel. This technique should be introduced to all eye care
specialists in order to effectively document retinal findings when
there are no retinal cameras available; this technique could also
be introduced to emergency room personnel in order to evaluate a
patient’s retina with minimal equipment cost.
Commercial Relationships: Luis Leon, None; Jonathan I.
Huz, None; Arkadiy Yadgarov, None; Marco A. Zarbin, None;
Neelakshi Bhagat, None
The Potential Positives of Smartphone Photography Using
Negative Images
David L. Nash, Eric Crouch, Shannon McCole. Ophthalmology,
Eastern Virginia Medical School, Norfolk, VA.
Purpose: Smartphone photography has become increasingly popular
and has several applications. We retrospectively reviewed and
compared standard color photographs with the negative photographic
image of several retinal findings photographed with a handheld
smartphone to identify any utility in the use of negative retinal
images with smartphone photography.
Methods: Using an iPhone 4S (Apple Inc., Cupertino, CA, USA)
and a handheld 20 or 28 diopter lens to capture retinal images of
4 separate patients with a variety of retinal conditions seen by a
single resident at a metropolitan based, residency program, color
photographs were compared to the negative images. The negative
images were created with the application Negative Me (Fuen Mao ©
2010-2013 1 stfancy Studio). The negative images were compared
retrospectively to their color counterparts. Photographs were cropped
to remove patient identifying information. All patients signed hospital
consent forms to be photographed.
Results: Smartphone negative retinal images, when compared to
their color image, best highlight hypopigmented lesions (such as
Cytomegalovirus (CMV) scarring), lightly colored structures on the
retina such as the optic disc, and areas or retinal pigment epithelium
loss. Examination of negative images taken three weeks apart of a
patient with CMV retinitis may highlight areas of retinal scarring.
Conclusions: Smartphone fundoscopy has many applications.
Negative images are simple to create using smartphone technology.
Certain conditions, most notably those involving hypopigmentation
of the retina, may benefit from the use of negative imagery. There
may also be utility in evaluating cup to disc ratios with negative
imaging. This study acts as a pilot study. Further study is needed to
validate the utility of negative imagery and its clinical application in
disease management.
Image 1: a: Color photograph taken with a 20 diopter lens. b: Color
photograph taken with a 28 diopter lens. c: Negative image of 1a. d:
Negative image of 1b. All images are inverted.
the corresponding standard fundus photos, only 28% of respondents
changed their initial diagnosis correctly. The iPhone camera was
easy to use and was able to image far peripheral lesions that were
located beyond the capture range of the stationary fundus camera. It
allowed documentation of a peripheral retinal lesion during dynamic
depression. It also has the advantage of being considerably less
expensive and more portable than other modalities.
Conclusions: The Smartphone-adapted indirect ophthalmoscope
can provide clinically useful imaging of the posterior pole and
retinal periphery. This efficient, cost-effective, and portable imaging
modality will allow rapid acquisition, storage, and transmission of
fundus images in the pediatric population and in patients unable to
sit for standard fundus photography. We believe it will be especially
useful for documenting and monitoring the fundus changes in
retinopathy of prematurity.
Commercial Relationships: Lianna Valdes, None; Dov B. Sebrow,
None; Tongalp H. Tezel, None
Support: Columbia University College of Physicians and Surgeons Scholarly Projects Program
Image 2: a: Color photograph taken with a 20 diopter lens. b: color
photograph taken 3-4 weeks later of the same eye with a 20 diopter
lens. c: Negative image of 3a. d: Negative image of 3b. All images
are inverted.
Commercial Relationships: David L. Nash, None; Eric Crouch,
None; Shannon McCole, None
Wide-Field Bedside Imaging of the Human Retina Using a
Smartphone Adapter
Lianna Valdes, Dov B. Sebrow, Tongalp H. Tezel. Edward S. Harkness
Eye Institute, Department of Ophthalmology, Columbia University
College of Physicians and Surgeons, New York, NY.
Purpose: To develop and evaluate the clinical utility of a
smartphone-based portable fundus camera that can be used for widefield imaging and recording of fundus pathologies at bedside.
Methods: A polyurethane mount was designed and constructed
to keep the camera module of a smartphone (iPhone 4S) aligned
with the assistant mirror of an indirect ophthalmoscope (Keeler
Inc., Broomall, PA). Pretesting was performed to determine the
optimal lighting and focusing conditions for video and single-shot
photography of the human fundus. Next, macular and peripheral
images of various fundus pathologies were obtained using both
a standard fundus camera (Zeiss FF450, Dublin, CA) and the
smartphone-based camera. The clarity of the images, correct
identification of the pathologies, and ease of obtaining and storing
the fundus images were compared using 8 masked ophthalmologists.
A 3D printable model of the camera mount was created with a 3D
scanner.
Results: High-quality imaging of the posterior pole and retinal
periphery was achieved using the novel iPhone-adapted indirect
ophthalmoscope. Best results were obtained when positioning
the iPhone camera 0.5 cm from the assistant mirror and using
medium intensity illumination (~0.6 mW cm-2 sr-1) from the indirect
ophthalmoscope. When looking at the fundus images captured by
the smartphone camera, survey participants accurately described
the details of the fundus pathology 72% of the time. After viewing
AMD - A Metamorphopsia Detector
Daniela Claessens1, Ronald V. Krüger2. 1Augenärztliche
Gemeinschaftspraxis Lindenthal, Cologne, Germany; 2app4eyes,
Meinerzhagen, Germany.
Purpose: Purpose: We developed the computer based App AMD
-A Metamorphopsia Detector- to provide a self-control monitoring
tool to detect and control the degree and size of metamorphopsias
and scotomas. This study was performed to examine sensitivity
and specificity of AMD and its correlation with anatomical course,
functional development and quality of life.
Methods: Methods: AMD is based on the Amsler grid. The
software uses the concept of a negative image: a distorted image
can be straightened by moving the mouse. Degree and dimension of
distorted lines or scotoma are transformed into indices.
I) In a pretest sensitivity and specificity of the AMD App were
examined in 17 healthy and 5 eyes with metamorphopsias due to
macular degeneration (age 61-83).
II) In a second step 19 eyes (age 24-91) with maculopathies
performed the metamorphopsia module alone or in combination with
the scotoma module.
III) In a prospective observational clinical pilot study we examined
monocular best corrected visual acuity (BCVA), Amsler grid,
vision related quality of life (National Eye Institute Visual Function
Questionnaire NEI VFQ 25), OCT (optical coherence tomography)
and AMD in 13 eyes (age 37-91) with macular edema before and after
intravitreal injections of anti vascular endothelial growth factor. Prior
to the study all patients signed informed consent.
Results: Results: I) Pretests confirmed construct validity and
reliability. Sensitivity of AMD was 80%, specificity was 100%.
Chi2-Test lead to rejection of the zero hypothesis (c2 > 3.84; freedom
degree = 1, α = 0.05).
II) Including a wider range of pathologies, addition of the scotoma
module increased sensitivity from 0.84 to 0.94. OR to detect macular
pathologies without metamorphopsias was 3.37; 95% CI: 0.94 - 5.67
when combining both app modules.
III) a-VEGF improved OCT and AMD in all 13 eyes: average
decrease of macular thickness was 77.54 mm (median 55 mm; SD
49.99; CI 104.75; 50.33) and 6.0 (median 4.76; SD 3.5; CI 7.9; 4.1)
for metamorphopsia index. Mean gain in NEI VFQ 25 was 8.77
(median 19; SD 12.39; CI 15.51; 2.03).BCVA improved in 10 and
was stable in 3 patients. Geometric mean BCVA was 0.38 logMAR
(SD 0.28) before and 0.25 logMAR (SD 0,21) after treatment.
Conclusions: Conclusions: As a quantitative test AMD can support
conventional diagnosis, lower the threshold for diagnosis,enhance
compliance and adherence in the treatment of age related macular
degeneration and other diseases leading to macular edema.
Commercial Relationships: Daniela Claessens, None; Ronald V.
Krüger, app4eyes (I)
Smart Ophthalmics: A Smart Service Platform for TeleOphthalmology
Wolfgang Fink, Mark Tarbell. Vis & Autonomous Explorat’n Sys,
University of Arizona, Tucson, AZ.
Purpose: To provide ophthalmic healthcare to people who are
geographically dispersed (e.g., rural populations), or operate/
live in austere environments (e.g., military, third world, natural
disaster), where time, cost, and possibility of travel make access to
even adequate medical care difficult. As a result, significant causes
of preventable vision loss, such as ocular trauma, glaucoma, and
macular degeneration, may be detected early and treated in time to
prevent permanent vision impairment.
Methods: Optical extension devices are attached to a smartphone to
turn it into a mobile ophthalmic examination device. A custom app
is engaged to perform an eye exam specific to the attached optical
extension. Using a wireless connection, the smartphone app submits
the collected examination data to a remote “expert system,” which
provides medical analysis. The analysis results are sent back to and
displayed on the smartphone.
Results: A server-based telediagnostic analysis capability for current
and future smartphone-based ophthalmic examination devices has
been developed. Examination data, gathered with such devices, can
now be sent via WiFi or cell signal through a smartphone app to a
server for automated analysis, the results of which are sent back to
the originating smartphone almost immediately. As a first analysis
example, the cup-to-disc ratio in fundus images is determined
to assist the early detection of glaucoma. The fundus images are
obtained via a commercial-off-the-shelf handheld ophthalmoscope
that is attached to a smartphone. Moreover, a smartphone-based
miniaturized pupillometer has been developed that allows for the
recording of pupillary behavior in the presence and absence of light
stimulation, i.e., pupillary light reflex and dark reaction. The pupillary
diameter as a function of time is displayed in addition to parameters
such as latency, constriction, and dilation times.
Conclusions: Smartphone-based mobile ophthalmic examination
devices combined with server-based smart telediagnosis
capabilities pave the way towards smart ophthalmics, i.e., smart
tele-ophthalmology. This paradigm may greatly improve remote
patient screening and triage. It may help ensure that patients with
undiagnosed eye diseases are detected early and treated in time to
prevent permanent vision impairment. Establishing this paradigm
has the potential to change the very economy of extending quality
healthcare to many while reducing cost.
Commercial Relationships: Wolfgang Fink, California Institute of
Technology (P), University of Arizona (P); Mark Tarbell, University
of Arizona (P)
Support: NSF Award #IIP-1430062
Mobile application software (App) for Prediction of Advanced
Age-related Macular Degeneration
Min-Lee Chang1, Yu-Hsuan Chiu3, Chung-Jung Chiu1, 2, Luca
Avoni4. 1USDA Human Nutrition Research Center on Aging, Tufts
University, Boston, MA; 2Department of Ophthalmology School of
Medicine, Tufts University, Boston, MA; 3Concord-Carlisle High
School, Concord, MA; 4Department of Ophthalmology, Bologna
Hospital Ospedale Maggiore, Bologna, Italy.
Purpose: To develop a state-of-the-art mobile application software
(App) for the macular risk scoring system (MRSS).
Methods: The MRSS algorithm uses information from a patient’s
age, sex, race, education level, smoking status, macular pigmentation
(normal vs. abnormal), maximal drusen size, and drusen texture (hard
vs. soft) to calculate a risk score for progression to advance agerelated macular degeneration (AMD). Based on a one-page prototype
of the MRSS App for the Age-Related Eye Disease Study (AREDS)
that have been developed and published along with the MRSS paper
(Chiu et al., Ophthalmology 2014;121:1421-1427.), we wrote a userfriendly App for iOS devices (iPhones/iPads) and Android devices
and submitted the App to the Apple App Store and the Google Play,
respectively.
Results: The App consists of 12 pages, including a cover page,
a welcome page, a credit page, 8 question pages for the eight
predictors, and a result page that automatically presents the result
from the MRSS. Before submitted for download, the App was tested
to ensure consistent results with both of the prototype of the App and
manual calculation from the MRSS algorithm. The App is free for
download at https://itunes.apple.com/us/app/macular-degenerationtest/id944048670?l=it&ls=1&mt=8 for iOS devices and at https://
play.google.com/store/apps/details?id=com.arscolor.app.mrss&hl=it
for Android devices.
Conclusions: We have refined the AREDS MRSS App prototype and
made the App available for download via Internet. To our knowledge,
our App is the first mobile application for the prediction of advanced
AMD. Based on the results from the MRSS paper, this App can serve
as a platform for modifications to work in other populations.
Commercial Relationships: Min-Lee Chang, None; Yu-Hsuan
Chiu, None; Chung-Jung Chiu, None; Luca Avoni, None
A New iPad and iPhone App to Manage Patients with Dry Eye
caterina cordella1, Edoardo Oliveri3, Massimo Sonnati3, Stefano
Barabino2. 1Farmigea SpA, Pisa, Italy; 2Clinica Oculistica,
DiNOGMI, Azienda Ospedaliera San Martino-IST, Genoa, Italy;
3
Hippocrates Sintech, Genoa, Italy.
Purpose: Dry eye syndrome (DES) is a disease of the ocular
surface and one of the most frequent pathological conditions in
ophthalmology. The aim of the project was to develop an app to offer
the possibility to use iPad and iPhone to record patients’ complete
clinical chart data with a specific section dedicated to ocular surface
tests and DES management that could be used in everyday clinical
practice and for clinical studies.
Methods: In the first phase we compared numerous clinical charts
available for PC and paper versions. Then we developed a complete
clinical chart (iEyes) to record patients’ picture, medical and
ophthalmological history, visual acuity, and clinical data for iPad.
In particular we have included a section with specific ocular surface
tests: dry eye questionnaires, Schirmer test, fluorescein and lissamine
green staining grading scales, tear break-up time, conjunctival
hyperemia score, eyelid margin redness, quality of expressed
meibum, and other specific tests used in experimental studies only.
The app was tested by 3 general ophthalmologists and 3 specialists in
ocular surface diseases. Data were recorded with standard methods
(paper) and with the new app. Recording time were measured and
compared. Data could visualized on i Phone also.
Results: A total of 300 patients were recorded, with ocular surface
tests performed in 90 patients. No problems were referred by
physicians with the use of the app on iPad. The recording time was
significantly (P<0.05) lower in the iPad group compared to the paper
group. The chance to visualize clinical chart recorded data on iPhone
obtained a significant satisfaction rate.
Conclusions: This new app provides a useful tool for clinicians and
researchers to record data that could be used in everyday practice
and for clinical studies. Future developments could be the possibility
for patients included in clinical studies to communicate data of
questionnaires from their devices and to verify treatment compliance.
Commercial Relationships: caterina cordella, Farmigea SpA
(E); Edoardo Oliveri, Hippocrates Sintech (E); Massimo Sonnati,
Hippocrates Sintech (E); Stefano Barabino, None
Automated ocular surface staining analysis based on a teachable
algorithm: smartphone web-based applicayion for dry eye and
ocular surface disease
Felipe H. Padovani1, Paulo Borges5, Kristopher Sanford5, Luciana
Oharomari4, Eduardo M. Rocha4, Paulo Schor3, Monica Alves2.
1
Pontific Catholic University of Campinas, Jundiaí, Brazil;
2
Ophthalmolgy, University of Campinas, Campinas, Brazil;
3
Ophthalmology, Federal University of São Paulo, São Paulo,
Brazil; 4Ophthalmology, FMRP, Ribeirão Preto, Brazil; 5Electrical
Engineering, Santa Clara University, Santa Clara, CA.
Purpose: Dry Eye Disease (DED) and ocular surface disorders
diagnosis, scoring and follow up apply vital stainings. However,
interpretation is influenced by observer experience and subjectivity.
The aim of the present work is to describe a web-based application
(app) which runs an algorithm to analyzes images of fluorescein
ocular surface staining obtained with the digital camera of a
smartphone to quantify, register and provide a follow up tool.
Methods: Support Vector Machine (SVM) was applied to score
the intensity and pattern of fluorescein staining in images of ocular
surface obtained from healthy individuals and ocular surface diseases
patients using the camera of the smartphone connected to a slit lamp.
We are trying an automated approach to reduce the noise and to crop
the unwanted background, using a sequence of algorithms from the
Open Computer Vision (OpenCV) programming library. The goal is
to obtain an image that can be analysed by a standard method which
produces a numerical coefficient tha can be compared to clinical
results.
Results: The software determines the optimal hyperplane to provide
the most accurate maximum margin to separate the staining area
from not one, as seen in punctate keratitis in DED, corneal ulcers and
epithelial defects. Gauss filters are being applied to improve the final
results. Preliminary results have showed necessity to improve the
techniques in order to reduce the noise and crop the image properly
and, then, improve the statistical and clinical relevance of the
numeric coefficient.
Conclusions: The present work describes a new tool capable to
register and score corneal fluorescein staining. It will be useful for
clinical standardization and follow up measurements of DED severity
and ocular surface disorders. Regarding practical aspect this tool can
run in devices present in the daily clinical practice, the slit lamp and
the smartphone, providing a feasible and more precise method to
analyse ocular surface staining.
Oculus Rift DK2 (Oculus VR, California) for stereoscopic viewing of
the recorded consultation.
Results: Five patients consulting for diseases of the anterior and
posterior segments were examined remotely and the diagnosis was
consistent with that of the first examiner in all cases.
Conclusions: The success of this experience of telemedicine and
augmented reality offers significant opportunities in the fields of
education and remote consultation.
Commercial Relationships: Mathieu Flores, None; Nina Mauris,
None; Anne-Charlotte Level, None; Julien Magnin-Feysot, None;
Maher Saleh, None
Normal ocular surface: fluorescein staining under cobalt filter,
smartphone and slit lamp photograph 10x magnification.
Image editing algorithm prior to staining analysis.
Commercial Relationships: Felipe H. Padovani, None; Paulo
Borges, None; Kristopher Sanford, None; Luciana Oharomari,
None; Eduardo M. Rocha, None; Paulo Schor, None; Monica
Alves, None
Contribution of augmented reality in clinical practice in
ophthalmology
Mathieu Flores1, Nina Mauris2, Anne-Charlotte Level2, Julien
Magnin-Feysot2, Maher Saleh1. 1CHU Besançon, Besançon, France;
2
ISIFC, Besançon, France.
Purpose: Feasibility study of a device allowing recording,
transmission and viewing by another operator of a stereoscopic
clinical examination at the slit lamp.
Methods: Two medical cameras (CM900, Haag-Streit, Bern,
Switzerland) were mounted on a slit lamp (BQ900, Haag-Streit,
Bern, Switzerland) equipped with a beam splitter allowing to record
stereoscopic videos. The video footage associated with the patient’s
clinical data were then remotely transmitted to another examiner
located in another room via a secure server dedicated to telemedicine.
The data was then visualized through augmented reality glasses type
A novel imaging technique to non-invasively track molecules in
the eye
Guillaume Normand1, Jan Penraat2, Joy Ghosh4, Kellyann Kovach2,
Mikele Simkins2, Michael Rosol1, Cynthia L. Grosskreutz3, Sudeep
Chandra1. 1BioMarker Development, Novartis Institutes for
Biomedical Research, East Hanover, NJ; 2LAS, Novartis Institutes
for Biomedical Research, East Hanover, NJ; 3Translational Medicine,
Novartis Institutes for Biomedical Research, Cambridge, MA;
4
Ophthalmology Research, Novartis Institutes for Biomedical
Research, Cambridge, MA.
Purpose: Understanding biodistribution in the eye is a critical step
in developing novel therapeutics, yet outside of small animals the
ability to visualize such distribution remains limited. A novel in vivo
imaging method was evaluated to non-invasively and longitudinally
measure the distribution of molecules after intravitreal (IVT)
injection. The goals of this study were to select the optimal dye
for in vivo use, to build a quantification method and to develop a
translatable technology.
Methods: The near-infrared probes IRDye800CW and Indocyanine
Green (ICG), as well as fluorescein, were evaluated ex vivo and in
vivo after IVT or intravenous (IV) dosing. Cynomolgus monkeys
were taken to the imaging room under anesthesia. After pupil
dilation, animals were placed in the prone position and images were
captured with a Spectralis, a confocal laser ophthalmoscope (cSLO),
using 30 degree and ultra-widefield lenses (UWF) and at different
sensitivities and diopters. Fluorescence intensity was measured at the
retina and in the vitreous using the CellProfiler software.
Results: IRDye800CW was found to be brighter both ex vivo and in
vivo and less prone to quenching at high concentrations as compared
to ICG. Moreover, IRDye800CW was shown to be cleared less
rapidly than fluorescein and ICG from the retina and blood pool
after IVT or IV injection, respectively. Based on a dose comparison
study, the minimum detectable concentration of IRDye800CW at the
retina was determined to be between 0.0349ug and 0.349ug. After
IVT injection of 3.49ug of IRDye800CW, fluorescence intensity at
the retina peaked 24h post-dose but was undetectable after 15 days
post-dose. Interestingly, IRDye8000CW was observed to accumulate
in the optic disc as well as form speckles along the retinal arteries.
The UWF lens was able to demonstrate the high variability of the
injection site and pattern by capturing a larger area of the retina.
Conclusions: Of the dyes tested, IRDye800CW was validated in
this study as the optimal dye for in vivo ocular imaging and potential
future clinical use. Distribution of the free dye was measured in
several animals in order to benchmark the passage of the free dye
as compared to other targeted probes. In contrast to other methods,
this is a non-invasive technique that combines qualitative as well as
semi-quantitative measurements of a molecule’s vitreous passage and
provides useful anatomical information, especially for the retina.
Commercial Relationships: Guillaume Normand, Novartis (E);
Jan Penraat, Novartis (E); Joy Ghosh, Novartis (E); Kellyann
Kovach, Novartis (E); Mikele Simkins, Novartis (E); Michael
Rosol, Novartis (E); Cynthia L. Grosskreutz, Novartis (E); Sudeep
Chandra, Novartis (E)
In vivo low light level mouse retinal SLO imaging: limits for
backscattered intensity and autofluorescense detection
Pengfei Zhang1, Azhar Zam1, Edward Pugh1, Robert J. Zawadzki1, 2.
1
Cell Biology and Human Anatomy, University of California, Davis,
Davis, CA; 2Dept. of Ophthalmology & Vision Science, University of
California, Davis, Sacramento, CA.
Purpose: To evaluate feasibility of a custom mouse scanning light
ophthalmoscope (SLO) for low light level (below 1mW) mouse
retinal imaging. Low light imaging is desired to reduce potential
deleterious effects of light-tissue interactions, including heating
and photochemical reactions (e.g., reactive oxygen species; retinoid
byproducts from pigment bleaching).
Methods: A custom widefield (~50 deg VA) high-resolution retinal
imaging SLO with a 0 Dpt contact lens (affixed to the SLO optics)
was used to simultaneously acquire both backscattered intensity as
well as autofluorescense signal from mouse retina. Different scanning
speeds and light levels were tested to find the minimum light power
that gives good quality images. 469 nm light from a Fianium sc400
Supercontinuum laser source was used for imaging and a 503 nm
long-pass filter was used to select retinal autofluorescense. The
SLO detection system consisted of a Hamamatsu PMT (H742220) for reflected light and H7422-40 for fluorescense and Femto
preamplifiers (107 gain), and the sampling rate is 500kHz. Custom
Labview acquisition software allowed selection of the sampling
region (ROI) and frame rate. Pigmented (C57BL/6) and albino
(BALB/cJ) mice from Jackson Laboratory were used as subjects.
During imaging, mice were anesthetized with 2-3% inhalational
isoflurane anesthetic. Pupils were dilated and cyclopleged with
Tropicamide and phenylephrine.
Results: Example images of mouse retina (BALB/cJ, 13 mo)
acquired simultaneously in reflectance and fluorescence channels
during a single imaging session with different light levels are
presented (Fig. 1). The data were acquired with pixel sample time of
6 μs and sampling density of 256 x 256 pixels, resulting in frame rate
of ~2Hz. Control gain voltage of the PMTs was adjusted to match the
dynamic range of the DAQ during imaging.
Conclusions: Low light level mouse retinal imaging is essential to
allow study of physiological processes of the retina in vivo without
adversely affecting retinal structure and function by the probing beam
itself. Our data suggest that low light level SLO is feasible for mouse
retinal imaging. In particular, light levels below 1mW can be used
during mouse positioning, keeping light exposure to minimum.
Fig.1 Single frame and 50 frames summed reflectance(REFL) and
autofluorescence(AF) images with different light levels.
Commercial Relationships: Pengfei Zhang, None; Azhar Zam,
None; Edward Pugh, None; Robert J. Zawadzki, None
Support: UC Davis Research in Science & Engineering (RISE)
grant, UC Davis National Eye Institute Core facilities grant.
Depot Indocyanine Green Dye for in vivo visualization of
infiltrating leukocytes
Dawn A. Sim1, 2, Colin J. Chu2, Michael B. Powner2, Senthil Selvam2,
Adnan Tufail1, 2, Catherine A. Egan1, James W. Bainbridge1, 2, Richard
W. Lee3, 1, Andrew D. Dick3, 1, Marcus Fruttiger2. 1NIHR Biomedical
Research Center, Moorfields Eye Hospital, London, United Kingdom;
2
UCL Institute of Ophthalmology, London, United Kingdom;
3
Academic unit of Ophthalmology, School of Clinical Sciences,
University of Bristol, Bristol, United Kingdom.
Purpose: To assess a novel method to label and image myeloid cells
infiltrating the mouse retina and choroid in vivo, using a single depot
injection of Indocyanine green dye (ICG).
Methods: We assess the use of depot ICG for the purpose of in
vivo cell labelling on three different ocular models of inflammation
and angiogenesis - endotoxin-induced uveitis (EIU), experimental
autoimmune uveoretinitis (EAU) and laser-induced choroidal
neovascularisation (CNV) model. A near-infrared scanning
ophthalmoscope was used for in vivo imaging of the eye and flow
cytometry was used on blood and spleen to assess the number and
phenotype of labelled cells. ICG was administered 72 hours prior
to the induction of inflammation to ensure clearance from systemic
circulation.
Results: We found that in vivo intravenous administration failed to
label any leukocytes, whereas depot injection, either intraperitoneal
or subcutaneous, was successful in labelling CD11b+ myeloid cells.
Progression of inflammation could be traced over a period of 14
days following a single depot injection of ICG. Additionally, flow
cytometric analysis revealed that the predominant population of cells
stained by ICG are circulating and splenic reservoir CD11b+ myeloid
cells.
Conclusions: The translation of this approach into clinical practice
would enable visualisation of immune cells in situ. This will not only
provide a greater understanding of pathogenesis, monitoring and
assessment of therapy in many human ocular diseases but also opens
the ability to image immunity live for neurodegenerative disorders,
cardiovascular disease and systemic immune mediated disorders.
Commercial Relationships: Dawn A. Sim, Allergan (F), UCL
Business (P); Colin J. Chu, None; Michael B. Powner, None;
Senthil Selvam, None; Adnan Tufail, UCL Business (P); Catherine
A. Egan, None; James W. Bainbridge, None; Richard W. Lee,
None; Andrew D. Dick, None; Marcus Fruttiger, UCL Business (P)
Exploring photoreceptor directionality using quadrant pupil
detection
Brian Vohnsen, Salihah Qaysi, Pedro Anjos, Denise Valente. School
of Physics, University College Dublin, Dublin, Ireland.
Purpose: Photoreceptors are endowed with directional properties
that are expressed by their angular sensitivity to incident light and
directionality in retinal imaging known as the Stiles-Crawford effect
of the first kind (SCE) and the optical SCE, respectively. Individual
photoreceptor tilt can be analyzed by moving the incident light across
the pupil when capturing retinal images. The purpose of this study is
to examine with numerical analysis and fundus imaging the viability
of an alternative quadrant detection scheme in which retinal images
are captured through four equal-sized pupil sectors from which
individual cone photoreceptor tilts can be derived.
Methods: A numerical analysis has been performed for directional
light scattering from photoreceptor cones in the retinal mosaic
modeled as tri-layered structures that represent the location of
mitochondria in the ellipsoid, the inner-outer segment junction and
the posterior outer segment tip [Vohnsen, Frontiers in Optics (2014)
FW5F.5]. Each outer segment has been added random tilt of varying
amplitude. Simulated cone mosaic images are calculated through 4
equal-sized sectors in the pupil plane. The numerical predictions are
compared with experimental quadrant parafoveal fundus images for
healthy subjects.
Results: The sectored quadrant pupil allows not only determination
of total intensity images but also direct determination of local cone
photoreceptor tilts (which may differ from their psychophysically
relevant tilts) in two orthogonal directions using the sectored
backscattered light intensity. It is found that the method is highly
suited to determine tilt without requiring displacement of the incident
light in the pupil plane whereby the analysis becomes simplified
although it requires proper centration with respect to the SCE peak
location. Retinal images in the healthy subjects show only small
photoreceptor tilts in the parafoveal region but confirm the potential
of the technique.
Conclusions: The numerical analysis shows that the quadrant pupil
detection is suitable for rapid detection of individual photoreceptor
tilts in the photoreceptor mosaic of less than 1 degree. With
proper optimization this may have clinical potential for analysing
perturbations caused by drusen and retinal disease that may alter the
pointing of the photoreceptors causing a resulting deterioration of
vision.
Commercial Relationships: Brian Vohnsen, None; Salihah Qaysi,
None; Pedro Anjos, None; Denise Valente, None
Support: SFI 08/IN.1/B2053
Intraocular scattering compensation in macular pigment density
measurement
Dimitrios Christaras, Harilaos Ginis, Alexandros Pennos, Pablo
Artal. Laboratorio de Optica, Universidad de Murcia, Murcia, Spain.
Purpose: Intraocular scattering reduces the contrast of fundus
images, especially in flood-illumination imaging systems. Since
macular pigment density (MPD) is often estimated from the analysis
of multispectral retinal images, this contrast reduction may affect the
results. We performed experiments to better understand the severity
of this problem and to overcome it by compensating scatter, thus
obtaining a more accurate estimate of the MPD.
Methods: A new multispectral fundus imaging system based on
the double pass principle, also capable of measuring intraocular
scattering at various wavelengths, was built. In addition, a
computational method to estimate MPD from retinal images after
compensation of light scattering was developed. The experimental
setup includes a xenon lamp filtered by band-pass filters and
homogenized by diffusers,a motorized iris conjugated to the subject’s
retina and an electron multiplying CCD camera. To avoid reflections
and backscattered light from the lens and the cornea, the imaging and
illumination arms are spatially separated by means of rectangular
apertures conjugated to the pupil plane. Intraocular scattering was
estimated using an optical integrator method by projection of uniform
disks ranging from 0.4 to 6 degrees radius (Ginis, et al (2012),
Journal of Vision, 12(3), 1–10). For the MPD calculation, a pair of
foveal images was recorded at two wavelengths, 460 and 550 nm.
These images were then corrected for the effect of scatter by using a
standard deconvolution technique and the scatter free value of MPD
was calculated according to a formula (Delori et al (2001), J. Opt.
Soc. Am. A, 18, 1212–1230).
Results: The complete procedure was successfully applied to 2 young
subjects, LH and DC, with no known eye pathology. The observed
maximum value of the MPD was 0.34 and 0.28 D.U. for DC and LH
respectively, whereas without any scatter compensation the respected
vales were 0.18 and 0.16 D.U., exhibiting an underestimation at the
order of 45% for the maximum value and 35% for the mean value of
the total MPD. The effect depends highly on the overall scattering but
also on the difference in scattering for blue and for green.
Conclusions: Initial results from two young, healthy subjects have
shown an important underestimation of the MPD when the effect of
scattering is not taken into account. This effect is expected to be more
important in elderly subjects with elevated intraocular straylight.
Commercial Relationships: Dimitrios Christaras, None; Harilaos
Ginis, None; Alexandros Pennos, None; Pablo Artal, None
Methods for an Enhanced 3D Interpretation of Autoradiographic
Ocular Drug Distribution Data
Julie E. Whitcomb1, Susan S. Lee1, Michael Robinson1, Jie Shen1,
Elizabeth Spencer3, Meagan Krueger3, Harvey Pollack2, Rex Moats2,
Yang Tang2, Mayssa Attar1. 1Allergan, Irvine, CA; 2Childrens Hospital
of Los Angeles, Los Angeles, CA; 3Covance, Madison, WI.
Purpose: Understanding drug distribution in the eye following
various routes of administration is important for optimizing drug
delivery. The physiologic and anatomic barriers that affect drug
distribution to the site of action can be studied with autoradiography
(ARL), which detects radiolabeled drug compounds on two
dimensional film, but lacks anatomical landmarks for spatial
reference. Herein, we describe a method to enhance ARL images for
3D interpretation.
Methods: A model radiolabeled compound (MW: 430 g/mol)
was dosed unilaterally via daily repeated topical dose and a single
intravitreal injection (n = 2 monkeys). The frozen heads were
sectioned into 43 transverse cross sections 20 mm thick through
the eye. Sections were mounted on a lexan plate, photographically
imaged (Figure 1a), then exposed on phosphor imaging screens
(Figure 1b) with blood standards for radioactivity quantitation. A
novel algorithm was designed to align the anatomical photographic
images to the ARL slides. Radioactivity concentration as a function
of image density was generated and the individual images stacked to
render a 3D volume.
Results: The 3D reconstruction provided an anatomically accurate
and quantitative representation of the 2D phosphor images. The blood
standards were used to translate the log-linear concentration of the
radioactivity to a quantitative heat map (Figure 1c) to interpretat the
dimensional gradient. Interpolation between the individual slices was
done to visually illustrate the regional drug distribution between the
two routes of administration (Figure 2). Overall the topical dosing
resulted in localized distribution in the anterior region; whereas, the
intravitreal dose was more dispersed throughout the entire eye.
Conclusions: Traditional ARL films are generally difficult to
interpret in a 3D space and the drug distribution is difficult to
interpret without any anatomic references. Quantitative 3D renderings
of drug distribution from 2D digital autoradiographic images is a
useful tool for understanding local drug delivery in relation to the
anatomical structures and can be applied to other labeling techniques
to better understand drug distribution.
Photographic (top), Autoradiographic (middle), and Quantitative
Heat Map (bottom) Images with Topical (left) and Intravitreal
(right) Administration
3D Rendered Images Comparing Topical (left) and Intravitreal
(right) Administration
Commercial Relationships: Julie E. Whitcomb, Allergan (E);
Susan S. Lee, Allergan (E); Michael Robinson, Allergan (E); Jie
Shen, Allergan (E); Elizabeth Spencer, Covance (E); Meagan
Krueger, Covance (E); Harvey Pollack, Allergan (C), Childrens
Hospital of Los Angeles (E); Rex Moats, Allergan (C), Childrens
Hospital of Los Angeles (E); Yang Tang, Allergan (C), Childrens
Hospital of Los Angeles (E); Mayssa Attar, Allergan (E)
Microfocus computed tomography as a new method for highresolution imaging of microstructures in the eye: a pilot study
Christian Enders1, Eva Braig2, Jens U. Werner1, Gerhard K. Lang1,
Gabriele E. Lang1, Kai H. Scherer2, Franz Pfeiffer2. 1Department of
Ophthalmology, University of Ulm, Ulm, Germany; 2Department of
Pyhsics and Institute of Medical Engineering, Technische Universität
München, Munich, Germany.
Purpose: Histopathological examination requires processing and
provides a two-dimensional (2D) view on the sections of a specimen.
The aim of this study is to investigate the structural characteristics
of the eye in an ex-vivo study using high-resolution microfocus
computed tomography (micro ct) without the need of processing and
providing three-dimensional (3D) non-destructive sectional images.
Methods: The study was approved by the institutional reviewboard,
and informed consent was obtained. A total of 4 specimens were
used. Among them were 3 enucleated human bulbi and 1 exenterated
swine eye. The 3 human specimens had underwent regular
histopathological examination. Diagnoses of the human specimens
were angle closure glaucoma, retinal detachment and malignant
melanoma. The swine eye had been used for surgical training and
a CyPass micro-stent® had been implanted. All specimens were
fixated in paraffin. We obtained images with a high-resolution micro
ct system (v|tome|x by GE®) for 3D computed tomography and 2D
non-destructive X-ray inspection. The maximum resolution is 2-20
micrometers depending on the sample size.
Results: The obtained images showed good correlation with
histopathologic findings. The maximum resolution in our specimens
is 20 micrometers. In the 2D-images retinal, choroidal, uveal,
corneal and lental tissues can be distinguished among others. The
configuration of the anatomical structures e.g. the chamber-angle
can be assessed without damaging the integrity of the specimen.
The exact position of a surgically applicated device like the CyPass
micro-stent® can be assessed faster and more accurate than with
standard histological examination. The 3D-images give a helpful
overview of the anatomical structures but cannot distinguish between
tissues as good as the 2D-images.
Conclusions: The results of this study indicate that ex-vivo micro ct
is a helpful research tool in the examination of enucleated bulbi with
excellent correlation to histo-pathologic findings. Although at present
not applicable in vivo because of the radiation load and sample size,
micro ct is a valuable complimentary tool in ex-vivo diagnostics
and in the development of new surgical techniques and ophthalmic
devices.
3D rendering of a swine eye with a micro-stent
orthogonal 2D slice of a a swine eye with a micro-stent
Commercial Relationships: Christian Enders, None; Eva Braig,
None; Jens U. Werner, None; Gerhard K. Lang, None; Gabriele
E. Lang, None; Kai H. Scherer, None; Franz Pfeiffer, None
Two-photon Signals From Label-free Human Epiretinal
Membranes
Juan M. Bueno1, Xavier Valldeperas2, Francisco J. Avila1, Raquel
Palacios1, Pablo Artal1. 1Laboratorio de Optica, Universidad de
Murcia, Murcia, Spain; 2Hospital Universitari Germans Trias,
Badalona, Spain.
Purpose: Epiretinal membranes (ERMs) are composed of neural
elements and different cellular and extracellular matrix (ECM)
components usually visualized using multi-labeling methods. We
have used two-photon microscopy to image non-stained human
ERMs to analyze the sources of two-photon excitation fluorescence
(TPEF) and second harmonic generation signal (SHG). This would
help to better understand the organization and structure of ERMs.
Methods: A custom-developed multiphoton microscope equipped
with adaptive optics (Bueno et al., J. Biomed. Opt. 2010) was used
to record pairs of TPEF-SHG images originated from the ERM
structures in a backward configuration. Specimens from ten patients
were successfully removed during vitrectomy and ERM peeling,
without the use of ophthalmic dyes. Immediately after surgery,
ERMs were fixated with a paraformaldehyde solution overnight and
flatmounted for two-photon microscopy imaging. Series of images
across the specimen and 3D stacks as a function of depth were
acquired for the different samples.
Results: Both TPEF and SHG signals were obtained from the ERM’s
samples, revealing different structures. SHG signals confirmed the
presence of filamentous structures corresponding to collagen fibers
and covering the specimen. TPEF signal was significantly higher
probably due to the presence of numerous types of cells (Muller,
glial…). This cellular autofluorescence is originated from the
mitochondrial reduced pyridine nucleotides NAD(P)H and flavin
compounds. Moreover, the elastin of the ECM might also contribute
to TPEF images.
Conclusions: Two-photon microscopy provides a non-invasive tool
to image unstained isolated ERMs. TPEF/SHG signal combination
revealed detailed information on the ERM morphology not available
with classical imaging techniques. Since staining procedures are not
required, the sample manipulation is reduced. Non-linear imaging
techniques are useful in the visualization and analysis of healthy
and pathological retinal and epiretinal tissues. Farther studies should
investigate the dependence between TPEF/SHG signals and ERM
etiology (idiopathic, diabetic or inflammatory).
Commercial Relationships: Juan M. Bueno, None; Xavier
Valldeperas, None; Francisco J. Avila, None; Raquel Palacios,
None; Pablo Artal, None
Support: Ministerio de Economía y Competitividad, Spain (grant
FIS2013-41237-R)
Noninvasive functional retinal imaging disrupted by
pharmacologic manipulation of neurovascular coupling pathways
Daniel Y. Tso1, 2, Momotaz Begum1, 2. 1Depts of Neurosurgery and
Ophthalmology, SUNY Upstate Medical Univ, Syracuse, NY; 2SUNY
Eye Institute, Syracuse, NY.
Purpose: Intrinsic signal optical imaging studies have shown the
existence of stimulus-evoked signals in the retina that are likely to be
of outer retinal origin. Additional studies have also shown that such
reflectance change signals are dominated by hemodynamics. The
linkage between this stimulus-driven retinal activity and the resulting
hemodynamic basis of the measured reflectance changes is a form of
neurovascular coupling. However the precise chain of mechanisms
in the retina whereby photoreceptor activation leads to a vascular
response is unknown. Most previous studies of neurovascular
coupling in the retina have been in vitro. In the present study, we
sought to dissect the role of several prospective signaling pathways
in the observed stimulus-evoked neurovascular responses in vivo
through intravitreal injections of selected agonists, antagonists and
blockers, while imaging these retinal intrinsic signals.
Methods: Animals and Preparation. Adult cats were anesthetized,
paralyzed and positioned in a stereotaxic. Using a modified fundus
camera, the retina was stimulated with visible (550nm) patterned
stimuli and illuminated in the near-infrared (700-900nm), while
intrinsic optical signals were recorded with a CCD camera.
Drug-injections. Intravitreal injections included suramin (purinergic
antagonist), HET0016 (20-HETE inhibitor), PPOH (EET inhibitor)
and NECA (adenosine agonist).
Results: Previous retinal imaging studies using intravitreal injections
of blockers of inner retinal function (e.g. TTX, PDA, APB) yielded
little impact on the retinal imaging signals. However in the present
study, each drug selected to interfere with a particular neurovascular
coupling pathway had profound impact on the observed retinal
imaging signals. In some cases, the sign of the observed retinal
imaging signal changed from negative to positive.
Conclusions: This novel functional retinal imaging technique
is shown to reveal outer retinal function through its impact on
hemodynamics via the neurovascular coupling pathways. These
studies help establish the chain of retinal events from light absorption
to observed changes in retinal reflectance in vivo. It is anticipated that
an understanding of the underlying neural, biophysical and signaling
mechanisms will help determine the utility of this method for studies
of retinal function, health and related clinical applications.
Commercial Relationships: Daniel Y. Tso, None; Momotaz
Begum, None
RetinAsk: A Queryable Atlas of the Retina
Jessica Taibl, Samir I. Sayegh. The EYE Center, Champaign, IL.
Purpose: To present the improved version of The Queryable Atlas
of the Retina that allows for queries based on partial information,
including retinal images, of an unknown retinal condition.
Methods: Since the original presentation of the Queryable Atlas
of the Retina (Sayegh et al, AAO 2000) a number of events have
occurred that led to a fundamental redesign and improvement of
the Atlas. 1) OCT has become a major tool of diagnosis of retinal
disease, 2) the speed and memory capacity of computers, as well
as speed of implementation of search algorithms, have continued
to rise exponentially and 3) new treatments paradigm and novel
surgical techniques have profoundly transformed retinal therapeutics.
These three aspects have been utilized to redesign the Atlas while
maintaining its fundamental queryable nature which clearly
distinguishes it from other more traditional approaches.
Results: The new design allows for rapid queries to be resolved
based on limited characterization or partial information. The search
“by field” and “by image” features have been vastly improved.
Conclusions: The Queryable Atlas of the Retina can be used as a
powerful diagnostic support tool as well as a valuable teaching tool in
residency and retinal fellowship programs.
Commercial Relationships: Jessica Taibl, None; Samir I. Sayegh,
None
Physical and Optical Properties of Rigid Gas Permeable Contact
Lenses Measured with Optical Coherence Tomography
Dean A. VanNasdale, Gang Huang, Nicky Lai, Marisa Ciamacca,
Amanda Eilerman, Aaron Zimmerman, Alex Nixon, Molly Smith.
Optometry, Ohio State Univ College of Optometry, Columbus, OH.
Purpose: To calibrate the axial and lateral dimensions of spectral
domain optical coherence tomography (SDOCT) images using rigid
gas permeable (RGP) contact lenses. To quantify the physical and
optical characteristics of RGP lenses using cross sectional OCT
imaging.
Methods: The parameters of 5 spherical RGP lenses of the same
design, index of refraction = 1.433, were manually verified. The
posterior side of each lens was mounted on a flat surface and imaged
using the Heidelberg Spectralis SDOCT system with Anterior
Segment Module (Heidelberg Engineering, Carlsbad, CA). The 30
deg image included the entire cross section of the central lens and
extended beyond the lens boundaries to included locations of the
mounting platform only, which was used as an axial reference. The
lateral and axial pixel dimensions were calculated and divided by
the known diameter of the contact lens and center thickness. The
index of refraction of the lens material was used to compensate
for the thickness changes induced in the cross sectional imaging.
Optical path length differences created by the thickness and index of
refraction change of the lenses were confirmed by measuring by the
displacement of the flat mounting surface under the lens, relative to
the undeviated surface imaged outside of the lens boundaries.
Results: Lateral and axial characteristics of the OCT images could
be successfully calibrated using the known parameters of the RGP
lenses. The coefficient of variation for distance conversions of the 5
lenses was less than 1% for each of the axial and lateral dimensions.
The displacement of the flat surface caused by the optical path length
difference induced by the contact lens was linearly proportional to the
thickness of the contact lens (r2 = 0.9736).
Conclusions: RGP lenses can be used to calibrate cross sectional
OCT images in both the lateral and axial dimensions. Physical
characteristics of RGP lenses can be extracted from cross sectional
OCT data. Optical properties of RGP lens can be assessed using
optical path length differences induced by the RGP thickness and
index of refraction changes.
Commercial Relationships: Dean A. VanNasdale, None; Gang
Huang, None; Nicky Lai, None; Marisa Ciamacca, None; Amanda
Eilerman, None; Aaron Zimmerman, None; Alex Nixon, None;
Molly Smith, None
432 Advanced ophthalmic imaging
Wednesday, May 06, 2015 11:00 AM–12:45 PM
2C/3C Mile High Blrm Paper Session
Program #/Board # Range: 4366–4371
Quantitative Ultrasonography of Vitreous Correlates with
Contrast Sensitivity and Visual Acuity
Kenneth M. Yee1, Jonathan Mamou2, Ronald H. Silverman2, 3,
Justin Nguyen1, Jeffrey A. Ketterling2, J Sebag1. 1VMR Institute for
Vitreous Macula Retina, Huntington Beach, CA; 2F.L. Lizzi Center
for Biomedical Engineering, Riverside Research, New York, NY;
3
Department of Ophthalmology, Columbia College of Physicians and
Surgeons, New York, NY.
Purpose: Vitreous floaters can significantly impact quality of life
when affecting contrast sensitivity (CS) and visual acuity (VA) but
there are no objective measures of vitreous structure for clinical
evaluation of floaters. This study developed quantitative analysis of
B-scan ultrasound (QUS) to measure vitreous echodensities as an
index of disease severity and to determine if QUS correlates with
functional deficits in CS and VA.
Methods: QUS reproducibility was tested in 10 eyes of 10 subjects.
35 eyes from 28 patients (56 + 18 yrs; 17 men, 11 females) with
subjective vitreous floaters were assessed by QUS using a customized
probe (15 MHz center frequency, 20 mm focal length, 7 mm aperture)
in three different scan orientations (longitudinal, LONG; transverse,
TRANS; and longitudinal to the macula, LMAC) yielding three
parameters (energy, E; mean amplitude, M; and percentage of the
vitreous body filled by echodensities, P50) from whole-central
vitreous. Statistical analyses evaluated E, M, and P50 correlations
with CS (%W, Freiburg Acuity Contrast Test) and VA (decimal of
Snellen ratio).
Results: QUS was highly reproducible in the LMAC (rICC > 0.90)
and LONG (rICC > 0.83) orientations. In 35 eyes with floaters,
CS ranged from 1.2 %W (normal) to 5.6 %W (worst) (mean = 2.7
+ 1.2 %W) and decimal VA ranged from 1.0 to 0.33 (mean VA =
0.76 + 0.19). Three QUS measures at each of two different scan
orientations correlated significantly with VA: E in the LMAC scan
orientation (R = -0.57, P<0.0004), E in the TRANS scan (R = -0.56,
P<0.0006), M in the LMAC scan (R = -0.60, P<0.0002; Fig 1) M in
the TRANS scan (R = -0.60, P<0.0002), p50 in the LMAC scan (R =
-0.44, P<0.008) and p50 in the TRANS scan orientation (R = -0.48,
P<0.004).
These same six QUS measures strongly correlated with CS: E in the
LMAC scan orientation (R=0.77, P<5.0 x 10-8), E in the TRANS scan
(R=0.73, P<6.7 x 10-7), M in the LMAC scan (R=0.76, P<10-8), M
in the TRANS scan (R=0.78, P<3.0 x 10-8; Fig 2), p50 in the LMAC
scan (R=0.68, P<7.3 x 10- 6) and p50 in the TRANS scan orientation
(R=0.71, P<1.8 x 10-6).
Conclusions: QUS objectively provides quantitative assessment
of vitreous structure across a wide range of vitreous echodensities.
QUS indices correlate with VA and CS suggesting that QUS could
provide a clinically useful index of vitreous structural abnormalities
in floaters and other conditions.
Commercial Relationships: Kenneth M. Yee, None; Jonathan
Mamou, None; Ronald H. Silverman, None; Justin Nguyen, None;
Jeffrey A. Ketterling, None; J Sebag, None
Optical coherence photoacoustic microscopy for in vivo
multimodal retinal imaging
Tan Liu1, Xiaojing Liu1, Rong Wen2, Yiwen Li2, Carmen A. Puliafito3,
Hao F. Zhang4, Shuliang Jiao1. 1Biomedical Engineering, Florida
International University, Miami, FL; 2Bascom Palmer Eye Institute,
University of Miami, Miami, FL; 3Keck School of Medicine,
University of Southern California, Los Angeles, CA; 4Biomedical
Engineering, Northwestern University, Evanston, IL.
Purpose: To develop a near-infrared (NIR) optical coherence
photoacoustic microscopy (OC-PAM) system, which can accomplish
optical coherence tomography (OCT) and photoacoustic microscopy
(PAM) simultaneously by using a single pulsed broadband nearinfrared light source. The ultimate goal of the study is to provide a
multimodal in vivo imaging tool that can provide comprehensive
and quantitative information about the absorption and scattering
properties of the retina.
Methods: OC-PAM offers a unique opportunity for studying
the scattering and absorption of biological tissues because the
simultaneously acquired OCT and PAM signals are generated by
the same group of photons. The acquired OCT and PAM images
are intrinsically registered in the lateral directions. To make this
technique more suitable for eye imaging, i.e more tolerable by human
subjects, we built a NIR OC-PAM by using an ultrafast Ti:sapphire
laser amplifier operating under an unseeded mode. The light source
had a center wavelength of 800 nm with a FWHM bandwidth of 30
nm. The OCT signal was detected in the spectral domain while the
photoacoustic signal is detected with an ultrasonic transducer place in
contact with the eyelid. The measured depth resolution of OCT and
PAM are 9.9 μm (in air) and 46 μm, respectively.
Results: To test the capabilities of the system on multimodal
ophthalmic imaging we imaged the retina of pigmented rats. The
OCT images showed the retinal structures with quality similar to
conventional OCT while the PAM images revealed the distribution
of absorbers in the retina, which were confirmed with histology to
melanin in the RPE and choroid. Choroidal vessels cast shadows in
PAM images. Since the absorption of hemoglobin is relatively weak
at around 800 nm, the NIR PAM signals are generated mainly from
melanin in the posterior segment of the eye.
Conclusions: We have demonstrated the feasibility of an OC-PAM
system working in the near-infrared. By using a single pulsed
broadband NIR light source, OC-PAM can image the scattering and
absorption contrasts simultaneously. This system can provide both
structural information with good depth resolution as conventional
OCT and melanin specific absorption contrast of the eye.
Commercial Relationships: Tan Liu, None; Xiaojing Liu, None;
Rong Wen, None; Yiwen Li, None; Carmen A. Puliafito, None;
Hao F. Zhang, None; Shuliang Jiao, None
Support: n/a
Quantitative oximetry and chromophore mapping in the rat
inner retina with visible light spectroscopic OCT
Vivek J. Srinivasan, Shau Poh Chong, Conrad W. Merkle, Conor
Leahy, Harsha Radhakrishnan. Biomedical Engineering, University
of California, Davis, Davis, CA.
Purpose: The purpose of this study is to test the hypothesis that by
1) changing the wavelength of Optical Coherence Tomography to
the visible wavelength range and 2) developing and validating novel
spectroscopic algorithms, reliable estimates of oxygenation, total
hemoglobin, oxyhemoglobin, and deoxyhemoglobin in the retina can
be achieved.
Methods: A high-speed visible-light spectral/Fourier domain OCT
system was constructed for in vivo imaging of rodents, using a
supercontinuum light source. An ex vivo calibration system was also
devised to validate both oxygenation and hematocrit measurements.
Whole blood flowing through FEP tubing was imaged and the
corresponding sO2 values were estimated. Long Evans rats (Charles
River Lab, MA) were used in the imaging studies. During the
experiment, the rat was supplied a mixture of isoflurane in oxygen
and medical air through a ventilating system. Imaging was centered
on the optic nerve and required approximately 10 minutes.
Results: The ex vivo validation experiments showed good agreement
between hemoglobin concentrations and oxygenations measured by
our spectroscopic algorithm and those determined by a centrifuge and
blood gas analyzer, respectively. Imaging results for saturation, total
hemoglobin, oxyhemoglobin, and deoxyhemoglobin are shown in the
Figure.
Conclusions: Here we introduce, validate, and demonstrate methods
for quantifying oxygenation and hemoglobin content in the inner
retinal vessels with spectroscopic OCT. When combined with flow,
these methods will enable metabolic imaging of the inner retina.
The figure shows quantification of chromophores in the rat retina
in an en face view. The saturation map shows clear distinctions
between arteries and veins (A). The map of the maximum of the
product of total hemoglobin concentration and distance shows larger
values in larger vessels (B). A map of the maximum of the product
of oxygenated hemoglobin concentration and distance shows that
veins and arteries contain oxyhemoglobin (C). By comparison, under
the given experimental conditions, most of the deoxyhemoglobin
is contained in veins (D). It should be noted that quantitative
measurements of chromophores can be achieved by integrating
the maps (B-D) in the transverse plane (x and y dimensions). All
maps were displayed using an en face alpha map based on the local
coefficients of determination at each transverse location, averaged
over depth.
Commercial Relationships: Vivek J. Srinivasan, Optovue (P);
Shau Poh Chong, None; Conrad W. Merkle, None; Conor Leahy,
None; Harsha Radhakrishnan, None
Support: We acknowledge support from the National Institutes
of Health (R00NS067050), the American Heart Association
(11IRG5440002), and the Glaucoma Research Foundation Catalyst
for a Cure.
Ex Vivo Hyperspectral Autofluorescence (AF) Imaging of Retinal
Pigment Epithelium (RPE) in Human Donor Eyes with Agerelated Macular Degeneration (AMD)
Tal Ben Ami1, Yuehong Tong1, Paul Sajda4, Zsolt Ablonczy2, Christine
A. Curcio3, Theodore Smith1, Thomas Ach3. 1Ophthalmology,
New York University School of Medicine, New York, NY;
2
Ophthalmology, Medical University of South Carolina, Charleston,
SC; 3Ophthalmology, University of Alabama at Birmingham School
of Medicine, Birmingham, AL; 4Biomedical Engineering, Columbia
University, New York, NY.
Purpose: We have shown how hyperspectral imaging can extract
individual signals of abundant fluorophores ex-vivo in normal RPE
[1]. Herein we analyze donor eyes with AMD, hypothesizing that the
spectral signatures would vary with perturbations in RPE physiology.
Methods: Hyperspectral AF images were captured from 7 locations
in 5 RPE/Bruch’s-membrane (BrM) flat-mounts from donor eyes with
early to late AMD. Imaging was performed at 2 excitation bands,
436-460 and 480-510 nm, with emission captured between 420 and
720 nm in 10 nm intervals using a Nuance FX camera.
Results: Gaussian mixture modeling and mathematical factorization
were applied to extract 1 BrM spectrum and 4 abundant emission
spectra from RPE organelles, the latter peaking at mean wavelengths
of 519±7, 574±8, 599±4, and 644±10 nm (436 nm excitation). The
519 nm peak was blue-shifted ~50 nm relative to the corresponding
signal from normal eyes [1]. Spatial abundance images showed
unique signals localized to RPE granule aggregates, melanosomes,
and basal laminar deposits (BLamD) (Figure).
Conclusions: In AMD, some AF signals from RPE become
spatially discrete [2], and a morphological diversity of RPE granule
populations is well demonstrated using hyperspectral imaging.
Differences in spectra and their spatial distributions between normal
and AMD eyes may extend our knowledge of RPE pathophysiology
in AMD.
1. Smith, Post, Johri, Lee, Ablonczy, Curcio, Ach, Sajda:
Hyperspectral signal recovery of unknown fluorophores in the RPE.
Biomed Opt Express 2014.
2. Ach, ARVO 2014
Hyperspectral images of an eye with geographic atrophy from an
81-year-old donor. RGB panel: Composite RGB AF image of an
atrophic area containing RPE cells (yellow) overlying BrM (green),
showing 2 RPE cells with prominent signal (gray arrows), granule
aggregates (white arrows), sub-RPE tissue including BrM and
persistent BLamD with brighter but cloudy, non-uniform AF (pink
arrows). Graph: The 5 spectra (C1 to C5) recovered at 436 nm
excitation, and their spatial abundance images (C2-C4 shown, warm
colors indicate greater abundance). C2 represents persistent BLamD
blocked by melanosomes, C3 shows cells and aggregates, and C4
is consistent with melanosomes. Overlay panel: Each component
is assigned its own color, with mixing occurring where components
co-localize.
Commercial Relationships: Tal Ben Ami, None; Yuehong Tong,
None; Paul Sajda, None; Zsolt Ablonczy, None; Christine A.
Curcio, None; Theodore Smith, Advanced Cell Technologies (C);
Thomas Ach, None
Support: National Institutes of Health/National Eye Institute grants
EY021470 (RTS), EY06109 (CC), unrestricted funds from Research
to Prevent Blindness (RTS, CC), and EyeSight Foundation of
Alabama (CC). German Research Foundation AC265/1-1 (TA).
Immuno-DNIRA (ImmunoD) is a novel imaging technique that
identifies ex vivo labelled macrophages in the eye with confocal
Scanning Laser Ophthalmoscopy (cSLO)
Natalie Pankova1, Huiyuan Liang2, Xu Zhao2, Hai Wang2, Shelley
R. Boyd1. 1Laboratory Medicine and Pathobiology, University of
Toronto/St. Michael’s Hospital, Toronto, ON, Canada; 2St Michael’s
Hospital, Toronto, ON, Canada.
Purpose: Phagocytic macrophages (Mφs) are found in human disease
and in pre-clinical models, but their spatial and temporal dynamics
cannot be adequately addressed in serial tissue specimens. We
previously developed Delayed Near-Infrared Analysis (DNIRA), a
technique by which retinal pigment epithelium (RPE) cells can be
viewed in vivo in the days following systemic indocyanine green
(ICG) injection using cSLO. The purpose of this study was (1) to
determine if Mφs can be safely labeled ex vivo with ICG and detected
in the rat eye in vivo by cSLO imaging following their systemic
injection, and (2) to determine their spatiotemporal change following
retinal damage.
Methods: Peritoneal Mφs from Sprague Dawley rats were incubated
in culture medium +/- ICG (0.25 mg/ml, 30 min), and viewed by
white light and fluorescent scanning at 790 nm. Cells were then
permeabilized with Triton-X100 to release the dye, and assayed by
790 nm spectroscopy. RPE and retinal damage was induced in rats by
systemic injection of sodium iodate (vs saline control), and 107 ICGlabelled cells were injected per animal via tail vein. Sequential cSLO
images were taken in the minutes and days following cell injection.
Results: After ICG incubation, Mφs remain viable, take on a green
colour and fluoresce. Cellular permeabilization reverses these
findings and releases dye into solution, indicating it was internalized
by the cells. Immediately following systemic injection of ICGlabelled Mφs, motile fluorescent spots can be detected in the fundus
using the ICG excitation/emission filters of the cSLO. In control
(saline) animals, cells are found almost exclusively within the retinal
and choroidal vasculature. In toxin-treated animals, cells are seen
both within the vasculature and extravasating into areas of outer
retinal damage.
Conclusions: These data demonstrate that Mφs can be safely labelled
in vitro with ICG, and following their systemic injection can be
viewed in vivo both in a healthy and damaged fundus. Labelled cells
preferentially localize to areas of outer retinal damage, suggesting
they may be recruited during disease. We have termed this novel
technique Immuno-DNIRA (ImmunoD). Future studies using
ImmunoD will investigate the correlation of Mφ dynamics with
disease progression.
Commercial Relationships: Natalie Pankova, None; Huiyuan
Liang, None; Xu Zhao, None; Hai Wang, None; Shelley R. Boyd,
Translatum Medicus Inc (I), Translatum Medicus Inc (P)
Support: Translatum Medicus Inc
Split-detector imaging reveals photoreceptors, outer nuclear
layer somata and horizontal cells without contrast agents in the
living mouse retina.
Andres Guevara-Torres1, 2, David R. Williams1, 2, Jesse B. Schallek2.
1
The Institute of Optics, University of Rochester, Rochester, NY;
2
Center of Visual Science, University of Rochester, Rochester, NY.
Purpose: Despite the micron-level resolution of adaptive optics
scanning light ophthalmoscopy (AOSLO), many cell classes in the
mammalian retina are challenging to image because they provide
weak endogenous optical contrast. Here we use split-detector imaging
(Scoles et al., 2014) to enhance the contrast of cell boundaries,
rendering them visible in high-resolution imaging.
Methods: Anesthetized C57BL/6 mice were imaged with an
AOSLO using near infrared light. The detection arm of the AOSLO
was modified by placing a knife edge at the retinal image plane to
direct the left and right half of the point spread function into two
photomultiplier tubes (PMTs). Simultaneously captured signals from
each PMT were subtracted to generate a differential contrast image.
AO was used for defocus control and the vascular layers were used to
confirm axial position.
Results: We resolved a variety of retinal cells that are invisible with
conventional confocal AOSLO without contrast agents. We observed
a monolayer of photoreceptor distal processes that showed a Yellot’s
ring peak at 23 cycles/degree (figure 1a). This corresponds to a
photoreceptor density of 477,000 cells/mm2, consistent with previous
reports of photoreceptor density in the mouse. When focusing
slightly more vitread, we could distinguish a multilayer arrangement
of photoreceptor somata within the outer nuclear layer (figure 1b).
Above this, at the boundary of the outer plexiform layer, we resolved
a sparse mosaic of cells (figure 1c) with a diameter of 11 ± 2 μm
(mean ± SD). This mosaic of cells counted by 4 individuals showed
a density of 1250 ± 270 cells/mm2 (mean ± SD, 0.25 mm2 analyzed
between 10-25 degrees from the optic disc), which is consistent with
previous histological reports of horizontal cell density in mouse and
more than an order of magnitude different than any other neural cell
class in the retina.
Conclusions: This approach reveals a variety of cell structures
that are hidden when imaged by confocal systems. The capability
to image horizontal cells and the photoreceptor somata enables
quantification of these cell classes over the course of disease.
Optimizing this contrast method in the mouse may instruct best
strategies for imaging transparent cells in the human retina.
Figure1 Split-detector images of a)mosaic of photoreceptor distal
processes b)photoreceptor somas c)horizontal cells
Commercial Relationships: Andres Guevara-Torres, Canon, Inc.
(F), University of Rochester (P); David R. Williams, Canon, Inc. (F),
Canon, Inc. (R), Polgenix, Inc. (F), University of Rochester (P); Jesse
B. Schallek, University of Rochester (P)
Support: F32 EY023496-01, BRP EY014375, P30 EY001319.
The Schmitt Program on Integrative Brain Research Postdoctoral
Fellowship, and Canon Inc.
503 Image processing, analysis and image based diagnosis
Thursday, May 07, 2015 8:30 AM–10:15 AM
Program #/Board # Range: 5247–5295/A0001–A0049
Contributing Section(s): Clinical/Epidemiologic Research, Eye
Movements/Strabismus/Amblyopia/Neuro-Ophthalmology, Low
Vision, Retina
Program Number: 5247 Poster Board Number: A0001
Inter- and Intra-observer agreement of Newly Developed
Software for Retinal Arteriolar Focal Narrowing Quantification
Alauddin Bhuiyan1, Pallab Roy2, Akter Hussain2, Kim Lee2,
Ecosse L. Lamoureux3, 5, Tien Y. Wong3, 5, Elsdon Storey4, Yogesan
Kanagasingam1, Rao Kotagiri2. 1Australian E-Health Research
Centre, Commonwealth Scientific &Industry Rsch Org, Floreat, WA,
Australia; 2The University of Melbourne, Melbourne, VIC, Australia;
3
Singapore Eye Research Institure, Singapore, Singapore; 4Monash
University, Melbourne, VIC, Australia; 5DUKE-NUS, Office of
Clinical Sciences, Singapore, Singapore, Singapore.
Purpose: To assess the accuracy and repeatability of novel computer
based semi-automated software for quantifying the severity of retinal
arteriolar focal narrowing (FN), a hypertensive retinopathy sign, from
colour retinal photographs.
Methods: 23 retinal photographs were randomly selected from the
Envision Study, a population-based study of persons aged 70 years
or over. Total 300 images were considered. The rest images were
excluded for not having FN point or being poor quality. A novel
computer assisted software was developed to quantify retinal FN
by selecting vessel region semi-automatically where FN exists.
The software then crops the FN region, calculates the vessel edges,
computes the widths and finds the FN points by finding the width’s
sudden changes. The software provides quantified FN as continuous
numerical values and the severity level as none, mild, moderate
and severe. The accuracy of the FN computation method was tested
against the manually measured FN severity level by expert grader
through the Spearman rank correlation. Inter-grader reliability
were assessed by measurements from two individual graders of the
same eye and intra-session repeatability was assessed by grading
the same image at two different times, using Intra-class Correlation
Coefficients (ICC).
Results: Accuracy measurements indicated a Spearman rank
correlation (ρ) of 0.88 (p<0.001) between software measured FN
values and expert graders’ ranking on FN severity. Between-graders
and intra-session ICC values for FN were 0.94 (95% Confidence
Interval [CI]: 0.85-0.97) and 0.93 (95% CI: 0.83-0.97) respectively,
which indicates high repeatability of the proposed software for FN
severity measurement.
Conclusions: A novel semi-automatic measurement system for
retinal arteriolar focal narrowing quantification is demonstrated
for accuracy and grading repeatability for the first time. A good
agreement with standard grading and excellent reproducibility,
indicating that quantitative assessment of retinal FN is possible.
Steps of of FN computation method (a) Region of interest
selection from original image (b) spatial normalization
(c) noise removal and vessel enhancement (d) detected edge
after applying canny edge detector (e) mapped vessel edge using
Dijkstra’s shortest path algorithm (f) mapped edge overlapped
on the ROI (g) computed vessel caliber or width for finding FN.
Commercial Relationships: Alauddin Bhuiyan, None; Pallab
Roy, None; Akter Hussain, None; Kim Lee, None; Ecosse L.
Lamoureux, None; Tien Y. Wong, None; Elsdon Storey, None;
Yogesan Kanagasingam, None; Rao Kotagiri, None
Bayesian Test-Retest Repeatability Tells You Want You Want To
Know In Equivalency Of Imaging Metrics
Natalie Hutchings1, Dylan Simpson2, Trefford L. Simpson1. 1School
of Optometry & Vision Science, University of Waterloo, Waterloo,
ON, Canada; 2Dept. of Political Science, Wilfrid Laurier University,
Waterloo, ON, Canada.
Purpose: To investigate Bayesian approaches to assessing test-retest
variabilty and compare the findings with standard method comparison
statistics.
Methods: A dataset of objective measures of tarsal conjunctival
roughness were used to evaluate the methods. Images of the tarsal
conjunctiva (n=29) with various amounts of unevenness in their
specular reflection (‘roughness’), were manually processed (NIH
ImageJ) twice by a technician naive to the experiment’s purpose.
Three of the 13 metrics of roughness obtained1 were evaluated (Ra, Rq
& Rt) using a Bayesian approach. Prior distributions of the mean, SD
and normality of the difference between the 1st and 2nd evaluation
were estimated as wide normal, wide uniform and wide exponential
respectively. Posterior distributions were derived from which a joint
distribution of effect sizes was generated. The highest density interval
(HDI) of the posterior distribution of effect size was compared to
a region of practical equivalence (ROPE; Cohen’s d=0.5 & 0.8).
The interpretation was compared to standard metrics in method
comparison of CCC (R; package epiR) and, mean difference and
limits of agreement (LOA).
Results: Bayes: The mean difference (modal SD) for each parameter
were Ra +1.1(1.9), Rq +0.5(2.0) and Rt -0.6(1.7). For a medium effect
size (d=0.5) 37%, 87% and 90% of the posterior distribution HDI lay
within the ROPE for Ra, Rq and Rt, respectively. The corresponding
data for a large effect size (d=0.8) were 86%, 99% and 99%. CCC &
LOA: The mean difference (SD) were Ra +1.2(2.0), Rq +0.6(2.5) and
Rt -3.32(15.9). Comparative outcomes for Rt are shown in Figure 1.
Conclusions: In test-retest repeatability, the intention is to
characterise the difference between repeat measures. Point estimates
of mean difference and LOA may give misleading results when
there are departures from concordance. Since the Bayesian approach
directly determines the parameter of interest and its varaiblity,
acceptance of ‘no difference’ and gives more insight into equivalency.
1
Chinga G et al. J Microsc. 2007;227:254-65.
Evaluation of Fundus Image Illumination Normalization in
Longitudinal Diabetic Retinopathy Screening
Kedir Adal1, 2, Peter G. Van Etten3, Jose P. Martinez3, Lucas J.
van Vliet2, Koenraad A. Vermeer1. 1Rotterdam Ophthalmic Inst,
Rotterdam, Netherlands; 2Imaging Physics, Delft University of
Technology, Delft, Netherlands; 3Rotterdam Eye Hospital, Rotterdam,
Netherlands.
Purpose: To evaluate if fundus images that are normalized for
variations in luminance and contrast improve the sensitivity of
diabetic retinopathy (DR) related lesion detection and the efficiency
of longitudinal screening for DR and diagnosis.
Methods: Four-field fundus image sets captured from either eye
of 30 diabetic patients who had retinal exams in 2012 and 2013 as
part of an ongoing diabetic retinopathy screening program in the
Rotterdam Eye Hospital were included. The fundus images were
normalized for local variations in illumination, yielding uniform
luminance and considerably increased contrast, especially near the
edges of the fundus images (Adal et. al. WBIR 2014). Image sets
of both exams were registered to a common coordinate system after
which color and normalized mosaics (fig. 1) were produced for each
retinal exam (Adal et. al., ARVO Abstract 444, 2014).
The fundus mosaics of each eye were shown to two expert graders
to examine early DR symptoms. DR-related retinal abnormalities
were then annotated on the color mosaics and, in a second round, on
the normalized fundus mosaics. The ground truth was defined as all
abnormalities that were annotated by both graders in either mosaic,
provided that at least one of the graders identified the abnormalities
in the color fundus mosaic.
Results: 30 color and 30 normalized fundus mosaics were
independently annotated for DR-related lesions and annotated by
two expert graders. The results are summarized in Table 1. A far
higher sensitivity was found for both experts for normalized mosaics
than for color mosaics (97.2% vs 62.5%), at the cost of a somewhat
higher number of false positives per eye (6.0 vs 0.7). On average, the
experts annotated 1.7 and 6.2 abnormalities per eye in the color and
normalized mosaics, respectively.
Conclusions: Normalization of fundus images improved the
sensitivity of longitudinal screening for diabetic retinopathy.
The normalized fundus image enhances the visibility of retinal
features and the detection of lesions such as microaneurysms and
hemorrhages.
FIgure 1: (Left) )Posterior distributions of the differences, SD of
the differences and effect size. The highest density interval (HDI)
indicates probable values for each parameter. The region of practical
equivalence (ROPE) is shown for Cohen’s d=0.8. (Right) Shows
evaluation 1 vs. evaluation 2 (top) and the corresponding mean/
difference plot (bottom).
Commercial Relationships: Natalie Hutchings, None; Dylan
Simpson, None; Trefford L. Simpson, None
Support: CFI Equipment Grant; NSERC (TS)
the graders found an abnormality on the color mosaic.
Commercial Relationships: Kedir Adal, None; Peter G. Van
Etten, None; Jose P. Martinez, None; Lucas J. van Vliet, None;
Koenraad A. Vermeer, None
Fig 1. (a) Example annotations in a color and a normalized fundus
mosaic. Some lesions are missed in the color images but are detected
in the normalized fundus images. (b) Example in which several
abnormalities were found on the normalized mosaic although none of
Non-mydriatic color fundus imaging with the Digital Light
Ophthalmoscope
Matthew S. Muller1, Thomas Gast1, 2, Jeffrey Clendenon1, Allen
W. Ingling1, Kenneth A. Stanfield1, Jason J. Green2, Karthikeyan
Baskaran2, Bryan Haggerty2, Todd Peabody2, Ann E. Elsner2, 1. 1Aeon
Imaging, LLC, Bloomington, IN; 2School of Optometry, Indiana
University, Bloomington, IN.
Purpose: To provide non-mydriatic confocal color fundus imaging
of sufficient quality for screening for diabetic retinopathy despite
dark fundus pigmentation, small pupil, high refractive error, or other
anterior segment issues.
Methods: Non-mydriatic color fundus images of 34 volunteers
(aged 39.2 ± 13.2 yr) were acquired using the Digital Light
Ophthalmoscope (DLO). 10 subjects had dark fundi and/or high
refractive errors, as well as other anterior segment issues. Unique
to retinal cameras, the Digital Light Ophthalmoscope (DLO) uses
a digital light projector with LED light sources to provide the
illumination for both confocal imaging and fixation stimuli. The DLO
projects a series of lines across the fundus that is synchronized to the
electronic rolling shutter read-out on a 2D CMOS sensor, providing
high contrast confocal imaging that is highly customizable through
software. Monochromatic 40 deg field images were acquired with
alternating red and green LED illumination at 14.3 Hz and overlayed
to present a pseudo-color image to the operator in real time. To
reduce pupil constriction and patient discomfort while maintaining
strong blood absorption, the green illumination was long-pass filtered
with a 570 nm filter, and a 1.5mm entrance pupil and time-averaged
power of <30 uW was used.
Results: The DLO provided gradable quality non-mydriatic fundus
images in all tested subjects, including those with dark fundi or
pupils < 2 mm, as judged by an EyePACS certified grader. The
use of long pass filters in the green LED permitted high contrast,
non-mydriatic images with illumination wavelengths >570 nm and
limited pupil constriction. Retinal vessels at the 4th branch or smaller,
as well as neovascularization in diabetic retinopathy, could be seen.
Hyperpigmentation was clearly seen both peripherally as bear tracks
and centrally at the posterior pole. The aperture width and color
balance can be adjusted to provide high contrast and yet relatively
uniform and natural color across the image.
Conclusions: The DLO provides confocal color fundus images
in real time without the use of short (< 570nm) wavelength light.
Despite recruiting an especially challenging population that included
dark fundi, small pupils, high refractive errors, and media issues,
we achieved a 100% success rate in obtaining gradable images for
screening.
Commercial Relationships: Matthew S. Muller, Aeon Imaging
LLC (F), Aeon Imaging LLC (I), Aeon Imaging LLC (P), Aeon
Imaging LLC (R); Thomas Gast, Aeon Imaging LLC (E), Aeon
Imaging LLC (R); Jeffrey Clendenon, Aeon Imaging LLC (E);
Allen W. Ingling, Aeon Imaging LLC (E), Aeon Imaging LLC (R);
Kenneth A. Stanfield, Aeon Imaging LLC (E); Jason J. Green,
None; Karthikeyan Baskaran, None; Bryan Haggerty, None;
Todd Peabody, None; Ann E. Elsner, Aeon Imaging LLC (I), Aeon
Imaging LLC (S)
Support: NEI Grant EY020017
Dealing with inter-expert variability in Retinopathy of
Prematurity through machine learning
Deniz Erdogmus1, Veronica Bolon-Canedo4, Esra Ataer-Cansizoglu1,
Jayashree Kalpathy-Cramer2, Oscar Fontenla-Romero4, Amparo
Alonso-Betanzos4, Michael F. Chiang3. 1Northeastern University,
Boston, MA; 2Massachusetts General Hospital, Charlestown, MA;
3
Oregon Health & Science University, Portland, OR; 4University of A
Coruña, A Coruña, Spain.
Purpose: Understanding the causes of disagreement among experts
in clinical decision making has been a challenge for decades. In
particular, diagnosis of retinopathy of prematurity (ROP) shows a
high amount of variability. Computer-based image analysis is one
approach to improving diagnostic variability. However, a critical
unanswered question is to understand discrepancies in the sets of
retinal vascular features considered by different experts during
diagnosis. We propose a methodology that makes use of machine
learning to understand the underlying causes of inter-expert
variability.
Methods: A set of 34 retinal images were diagnosed by 22
independent experts. Feature selection (FS) is applied to discover
the most important features considered by a given expert. These
are compared in turn with those of the rest of experts by applying
similarity measures. Finally, an automated classification system with
the most relevant features is built to check if this approach can be
helpful in ROP diagnosis.
Results: The experimental results reveal that the top selected
features regardless of the considered expert are: mean of venous and
arterial tortuosity (for 100% and 47% of experts), mean of venous
acceleration (42% of experts), and maximum main branch leaf node
factor in arteries (68% or experts). For pairs of experts with high
percentage of inter-agreement, the FS methods also select similar
features. These findings suggest that besides taking into account
the standard features (arterial tortuosity and venous dilation), the
experts may be considering other features, and that this may be a
source of disagreement. Finally, we built an automatic system using
the relevant selected features, with which the classification accuracy
was improved from 68% to 80% when distinguishing plus, pre-plus
and neither; and maintained when classifying into plus or not plus,
showing 88% accuracy. The high Williams’ indices obtained by our
system (greater than 1) reinforce the idea that it shows a behavior
similar to that of expert clinicians.
Conclusions: We provide a handy framework to identify important
features for experts and check whether selected features reflect the
pairwise disagreements. These findings may lead to improved ROP
diagnostic accuracy and standardization among clinicians, and may
be generalizable to other clinical problems.
Commercial Relationships: Deniz Erdogmus, None; Veronica
Bolon-Canedo, None; Esra Ataer-Cansizoglu, None; Jayashree
Kalpathy-Cramer, None; Oscar Fontenla-Romero, None; Amparo
Alonso-Betanzos, None; Michael F. Chiang, Scientific Advisory
Board for Clarity Medical Systems (S)
Support: NIH grant EY19474 and grant EY22387, and Research to
Prevent Blindness
A GMM-based Feature Extraction Technique for the Automated
Diagnosis of Retinopathy of Prematurity
Esra Ataer-Cansizoglu1, Veronica Bolon-Canedo2, Deniz Erdogmus1,
Katherine Abrahams4, Susan Ostmo4, Robison V. Chan5, Jayashree
Kalpathy-Cramer3, Michael F. Chiang4. 1Electrical and Computer
Engineering, Northeastern University, Malden, MA; 2Computer
Science, Universidade da Coruna, A Coruna, Spain; 3Massachusetts
General Hospital, Charlestown, MA; 4Oregon Health & Science
University, Oregon, OR; 5Weill Cornell Medical College, New York,
NY.
Purpose: Plus disease is an important factor for identification of
severe retinopathy of prematurity. Arterial tortuosity and venous
dilation in the retina are important signs for determining plus disease.
In order to build a computer-aided diagnosis system, it is necessary
to extract these features from vessel points or segments. Then,
an image is represented with “regular” statistics (e.g. minimum,
maximum or mean of these values). However, these statistics provide
biased estimates as an image contains both healthy and abnormal
vessels. Moreover, one should distinguish between arteries and veins
to account severity properly. We have developed computer-based
image analysis methods to (1) overcome these limitations with a
novel feature extraction method that represents each image with the
parameters of a two-component Gaussian Mixture Model (GMM),
(2) compare the classification accuracy for the features extracted from
the whole image versus features extracted from arteries and veins
separately.
Methods: 77 retinal images were manually segmented by an expert,
and arteries and veins were manually annotated. Reference standard
diagnoses were provided for each image by a consensus of 3 experts,
together with the actual clinical diagnosis. Manual segmentations
were fed into our image processing system that yielded the proposed
GMM statistics and regular statistics of 10 image-based features.
We built an automatic diagnosis system for identification of plus
disease vs. preplus vs. neither with various classifiers using (i) GMM
statistics, (ii) regular statistics, and (iii) both of them on features
extracted from the whole image and on features extracted from
arteries and veins separately.
Results: The classifiers trained with the proposed features
outperformed the classifiers using regular statistical features. The
best classification accuracy among different classifiers was 90.5%
with the proposed features and 88.2% with the regular statistical
features. Extracting features from the whole image without
distinguishing veins and arteries gave better accuracy (90.5%)
compared to using features from arteries and veins separately
(80.2%).
Conclusions: Overall tortuosity of an image can be well represented
with a two-component GMM of the image-based features. GMM
statistics are useful for building accurate automated diagnosis
systems and they remove the necessity of distinguishing arteries from
veins.
Commercial Relationships: Esra Ataer-Cansizoglu, None;
Veronica Bolon-Canedo, None; Deniz Erdogmus, None; Katherine
Abrahams, None; Susan Ostmo, None; Robison V. Chan, None;
Jayashree Kalpathy-Cramer, None; Michael F. Chiang, Scientific
Advisory Board for Clarity Medical Systems (unpaid member) (S)
Support: NIH grant EY19474 and grant EY22387, St. Giles
Foundation, Research to Prevent Blindness
Rapid grading of fundus photographs for diabetic retinopathy
using crowdsourcing: External Validation
Christopher J. Brady1, Andrea C. Villanti2, Jennifer L. Pearson2,
Thomas R. Kirchner2, Chirag Shah3, Omesh P. Gupta4. 1Wilmer Eye
Institute, Johns Hopkins University, School of Medicine, Baltimore,
MD; 2The Schroeder Institute, Legacy, Washington, DC; 3Retina
Division, Ophthalmic Consultants of Boston, Boston, MA; 4Mid
Atlantic Retina, Wills Eye Hospital, Philadelphia, PA.
Purpose: To refine and externally validate a novel method for fundus
photograph grading.
Methods: A crowd-sourcing interface for fundus photo classification
developed for Amazon Mechanical Turk (AMT), including annotated
training images was refined based on user feedback. In Phase 1,
nineteen expert-graded images were posted for categorization into 4
severity categories by AMT workers (Turkers), with 10 repetitions
per photo. Three sequential batches were posted with iterative
refinements to the interface. In Phase 2, 400 images from the
MESSIDOR public datastet of non-mydriatic fundus photos were
posted using the refined interface from Phase 1, asking Turkers to
categorize the images as normal or abnormal. In Phase 3, iterative
improvements were made to the interface in an attempt to further
refine accuracy using the Messidor dataset. The main outcome
measure was proportion of images with matching consensus Turker
and expert/gold-standard score.
Results: Across 190 grading instances in Phase I, Turker consensus
accuracy in 4-category grading increased to a maximum of 52.6%
from 26.3%. Turker accuracy at categorizing the images as normal vs.
abnormal increased to 100% from a baseline of 89.5%. Throughout,
100% sensitivity for normal vs. abnormal was maintained. Maximum
specificity was 85.7%. Across 4000 grading instances in Phase 2,
Turkers had an overall accuracy of 68.5%. Excluding the first two
MESSIDOR disease categories, level 1 (<5 microaneurysms (MA))
and level 2 (<15 MA or <5 hemorrhages), accuracy increased to
80.9% with a sensitivity of 92.4% and specificity of 78.0%. Four
out of 53 cases (7.5%) of level 3 (≥15 MA or ≥5 hemorrhages or
neovascularization) retinopathy were missed.
Conclusions: With minimal training, the AMT workforce can
rapidly and correctly categorize fundus photos of diabetic patients as
normal or abnormal when a moderate to severe amount of disease is
present. Further refinement is required for Turkers to identify subtle
disease, and correctly categorize the level of disease. That Turker
accuracy was preserved using a different dataset than that with which
the interface was developed is a critical validation. Images were
interpreted for a total cost of $1.10 per eye. Crowdsourcing may offer
a novel and inexpensive means to reduce the skilled grader burden
and increase screening for diabetic retinopathy in some settings.
Commercial Relationships: Christopher J. Brady, None;
Andrea C. Villanti, None; Jennifer L. Pearson, None; Thomas R.
Kirchner, None; Chirag Shah, None; Omesh P. Gupta, None
Automated and Manual Non Mydriatic Digital Retinal Imaging
in Community-based Tele-Ocular Screening
Ann M. John, Sumana S. Kommana, Nicole Mendez, Saysha Blazier,
Bernard C. Szirth, Albert S. Khouri. Ophthalmology, Rutgers New
Jersey Medical School, Newark, NJ.
Purpose: Tele-ocular screening relies on accurate interpretation of
captured images. A newer automated fundus imaging system allows
fast image acquisition with minimal training of screeners. Additional
image enhancement assists in the ability to capture retinal pathology.
The purpose of this study is to compare the quality of images
captured by automated, enhanced automated, and traditionally used
manual non mydriatic retinal camera systems.
Methods: We prospectively enrolled 44 subjects (88 eyes) during a
community-based screening to be imaged with two fundus camera
systems: 1. Manual imaging (Canon CR-DGi- Tokyo, Japan) and
2. A novel automated system (Canon CR-2AF- Tokyo, Japan). All
subjects were imaged with ambient room light (250 Lux) without
receiving mydriatic agents. Time to image both eyes included time
to focus, capture in one eye, allow visual recovery of the second eye,
and capture in the second eye. Image quality was assessed using
a previously published validated quantitative grading scale (1-5,
5=best, 1=worse) (Table 1). An optometrist and ophthalmologist each
evaluated 87 images from the manual, 88 images from the automated,
and 88 images from the enhanced automated groups. Ratings were
compared using t-tests. Inter-rater reliability was measured using
Cohen’s Kappa (K).
Results: Subject characteristics were: 60.4% female, 39.6% male,
45.8% African-American, 50.0% Hispanic, 4.2% White. Mean
quality scores were 4.26±0.9, 4.50±0.8, and 4.30±0.8 in the manual,
automated, and enhanced automated groups, respectively. Time to
image both eyes was lower with the automated system (Table 2).
There was statistically significant higher grading in the automated
imaging system when compared to both the manual system and
enhanced automated system (p<0.05). The enhanced automated
image grading was higher than the manual image grading but
did not reach statistical significance (p=0.33). Overall inter-rater
reliability was moderate (mean K=0.49, upper limit=0.59 and lower
limit=0.40).
Conclusions: In this sample population, images captured with a new
automated fundus imaging system had a faster acquisition and higher
quality score than manual imaging. The auto-enhance feature did
not increase image quality. In the future, automated imaging during
community screening of a larger population with diverse pathology
will be investigated.
1: Grading Scale (Kolomeyer et al., 2014)
2: Comparison of Cameras
Commercial Relationships: Ann M. John, None; Sumana S.
Kommana, None; Nicole Mendez, None; Saysha Blazier, None;
Bernard C. Szirth, None; Albert S. Khouri, None
Photograph Reading Center Front-end Automated Fundus Image
Analysis for Clinical Trials
Brian Madow. Ophthalmology, University of South Florida, Tampa,
FL.
Purpose: To design and implement automated query analysis of the
submitted fundus photograph images to the Photograph Reading
Center (PRC) for clinical trials by evaluating the characteristics of
the digital data and perform automatic check against the required
parameters. Currently the submitted images to the PRC are evaluated
manually, which is time and labor consuming process. We aimed
to increase the accuracy, decrease the time for the processing and
increase the effectiveness at the busy PRC.
Methods: Fast programing languages were used to develop
algorithms and friendly graphical user interface (GUI). The software
was tested on computer with Pentium quad processor running
Windows 7 operating system. Small set of 25 digital images was
initially used to test the principle and adjust the algorithms. Set of
100 consecutively received fundus photographic images at the PRC
were evaluated for accepted file format, file location, compression,
file size, resolution, image name, identifiable subject information,
image quality, and image color profile. Additional algorithm was
developed to extract and display the coded into the file information
about the name of the study, the clinical site, the subject code, the
visit number, the image type, the eye (right or left) and the date when
the image was taken. Supplementary module was added to allow for
operator input for other image characteristics pertinent to the clinical
trial that cannot be automated.
Results: The software performance was checked for accuracy against
manually evaluating the images by an operator. The test results
yielded 100 % accuracy from automated output. Images were retested
repeatedly 3 times and the accuracy was found to be reproducible
100%. The time for performing the automated testing was found to be
less than 1 sec., which is impossible to achieve by manual evaluation
of the images. The extracted information from the file was found to
be 100% accurate, when encoded as required.
Conclusions: Robust, reliable, automated software platform
was created, evaluated, validated and implemented at the Uveitis
Reading Center for assessment of vitreous haze grade from fundus
photographs. The software is applicable to fundus images of any type
and can be employed for use in other Reading centers or for other
studies of fundus images.
Commercial Relationships: Brian Madow, None
Development of a Focus Stacking Algorithm for Improvement of
3D Retinal Image Reconstruction
Luis A. Carvalho, Danilo Motta, Valeria Carvalho. Departamento of
Research and Development, Wavetek Technolgies/Universidade de
Sao Paulo, Sao Carlos, Brazil.
Purpose: There is an urgent need in ophthalmology for more cost
effective instrumentation for the early diagnosis of Glaucoma, a
leading disease in the cause of blindness worldwide. In a previous
project we have developed a low cost 3D retinal camera [1], which
presented promising results. Nevertheless, one of the great challenges
in 3D reconstruction is the precision with which the cross correlation
of stereo images is implemented, where in turn the depth of field
(DOF) also plays an important role.
Methods: In this work we have applied a focus stacking technique
to images acquired with a 3D retinal camera developed in a previous
project [1]. Stereo videos were acquired from a model eye for
progressive focusing planes of the object conjugated with a CCD
plane with zoom lens. A matlab algorithm was developed in order to
pre-process the video files and select specific image couples that had
very distinct focus characteristics. In-focus regions of each image
were detected automatically via edge detection and Fourier analysis
implemented in our in-house matlab algorithm. The in-focus patches
were then blended together to generate the final image.
Results: Our stacking algorithm was tested on a model eye for 3
distinct optic disc shapes manufactured with precision of 5 microns.
Results for the same algorithm developed in the previous phase [1]
were compared both for non focus stacking images and also after the
stacking procedure. The RMSE for all model eyes without stacking
was 0.14 mm and with stacking was 0.10 mm, almost 30% more
precise.
Conclusions: From the results obtained for this preliminary study on
model eyes the conclusion is that the stacking technique, originally
used in photography and microscopy, can be applied to 3D optic disc
reconstruction as one more factor for precision improvement. Our
goal in the next phase is to apply the same algorithms developed here
to in vivo eyes and compare results with other more sophisticated
(and more expensive) optic nerve depth analysis techniques, such as
SLO.
[1] CARVALHO LAV, RAMOS E, ROMANO A, Construction of a
True-depth Quantitative and Low cost 3d Optical Disc Measurement
Device and Collaborative Diagnosis using Stereo Images and Videos.
In: ARVO, Invest Ophthalmol Vis Sci 2012; 53: E-Abstract 3102,
2012.
Commercial Relationships: Luis A. Carvalho, Wavetek LLC (I);
Danilo Motta, Wavetek Brazil (C); Valeria Carvalho, Wavetek USA
(E)
Support: Wavetek LLC USA Grant
3D Reconstruction of Ocular Structures with Quadroscopic
Imaging Platform
Les Bogdanowicz1, John R. Hetling1, 2. 1Department of
Bioengineering, University of Illinois at Chicago, Chicago, IL;
2
Ophthalmology and Visual Sciences, University of Illinois at
Chicago, Chicago, IL.
Purpose: There is a need to improve access to ophthalmic
diagnostics in underserved populations, in particular for diagnosing
and monitoring glaucomatous optic neuropathy. This study
demonstrates proof of concept for a quadroscopic camera system
utilizing smartphone-like cameras for volumetric imaging of the
retinal surface.
Methods: Four high resolution cameras were arranged to image
intersecting planes. A model eye with three-dimensional features
for the optic nerve head and surrounding capillaries was designed
in Siemens PLM CAD software and 3D printed on an Objet350V
Pro printer using Verowhite. The lens is purchased from GWB
International, it is an optically lumped single lens mydriatic eye
model with a 6 mm pupil.
The camera system is a four-angle fundus camera, modeled in Zemax
and prototyped on an optical bench. The imaging system for each
camera has three effective lenses: an objective creates an intermediate
image of the retina, intermediate lenses to relay the image and
compensate for defocus of the eye, and a final set of lenses to focus
the image on the camera sensor. The cameras are mounted to a 3 inch
plate yielding a seven degree angle relative to the optical center of the
system.
The cameras are 5MP Ximea MU9PM-MH subminiature USB 2.0
cameras. Cameras were calibrated for focus utilizing a calibration
slide placed at the location of the retinal surface relative to the lens.
Cameras were connected via USB to a computer and images were
captured and processed in Matlab.
The Matlab algorithm provides camera calibration, registration, depth
recovery and 3D ocular image reconstructions from the quadroscopic
image. The resulting reconstruction were compared to the known
features in the Objet printed eye model.
Results: The algorithm calculated the height and depth of the optic
nerve head, and geometry of the capillaries, with an initial accuracy
of +/- 25 microns. This is lower resolution than provided by typical
OCT systems, but sufficient to reveal clinically relevant changes in
cup shape.
Conclusions: A quadroscopic camera system utilizing smartphonelike cameras for the reconstruction of 3D volumes of the retinal
surface is feasible. The next phase of development will include
evaluating the prototype system in human eyes for direct comparison
to OCT. The ultimate vision for this technology is a low-cost, easily
implemented diagnostic instrument to provide accurate volumetric
imaging of the retinal surface.
Commercial Relationships: Les Bogdanowicz, Ocumed (C); John
R. Hetling, None
Automatic quantification of geographic atrophy in
autofluorescence images of Stargardt patients
Clara I. Sanchez1, 2, Stanley Lambertus2, Bart Bloemen1, Nathalie
Bax2, Freerk G. Venhuizen1, Mark J. Van Grinsven1, Bram V.
Ginneken1, Thomas Theelen2, Carel C. Hoyng2. 1Diagnostic
Image Analysis Group, RadboudUMC, Nijmegen, Netherlands;
2
Ophthalmology, RadboudUMC, Nijmegen, Netherlands.
Purpose: To evaluate an observer-independent image analysis
algorithm that automatically quantifies the area of geographic atrophy
in fundus autofluorescence images of Stargardt patients.
Methods: Fundus autofluorescence images of 20 eyes of 20 Stargardt
patients with presence of one delineated or patchy atrophy region
in the macular area were selected. An image analysis algorithm was
developed to automatically segment the area of atrophy starting
from an arbitrarily selected seed point inside the atrophy region. The
method was based on a combination of region growing algorithm
and a dynamic, user-independent threshold selection procedure
using Otsu thresholding. In order to assess the performance obtained
by the proposed algorithm, manual annotations were made by an
experienced human grader. The grader manually delineated the
atrophy areas on the same set of images using dedicated software
developed for this task.
Results: A high correlation was observed between the manual area
measurements and the automatically quantified values obtained by the
proposed algorithm, with a mean intra-class correlation coefficient
(ICC) value larger than 0.89. In addition, the quantification time
was reduced substantially by a factor of 27 compared to manual
assessment. The output of the software was also shown to be
independent of the user input and highly reproducible, with an ICC
value larger than 0.99 between two executions of the algorithm at
different time points and with different seed points.
Conclusions: An image analysis algorithm for automatic
quantification of geographic atrophy in autofluorescence images of
Stargardt patients was developed. The proposed algorithm allows
for precise, reproducible and fast quantification of the atrophy area,
providing an accurate procedure to measure disease progression and
assess potential therapies in large dataset analyses independent of
human observers.
Figure 1. Original fundus autofluorescence image of a Stargardt
patient (left), manual annotation of the atrophy region (center) and
automatically quantified atrophy region obtained by the proposed
algorithm (right)
Commercial Relationships: Clara I. Sanchez, None; Stanley
Lambertus, None; Bart Bloemen, None; Nathalie Bax, None;
Freerk G. Venhuizen, None; Mark J. Van Grinsven, None; Bram
V. Ginneken, None; Thomas Theelen, None; Carel C. Hoyng,
None
Evaluating Software-Assisted Grading Performance In The
Detection Of Retinal Vasculature Abnormalities
Carla Agurto Rios1, E Simon Barriga1, Sheila C. Nemeth1, Elizabeth
McGrew1, Cesar Carranza1, Dalia C. Guadarrama Vallejo2, ETHEL
B. GUINTO ARCOS2, Peter Soliz1, Vinayak S. Joshi1. 1VisionQuest
Biomedical LLC, Albuquerque, NM; 2Instituto de Oftalmologia
Fundacion Conde de Valenciana, Mexico City, Mexico.
Purpose: To evaluate the effectiveness of the comprehensive
assessment of retinal vasculature (CARV) software as a screening aidtool for readers in the detection of retinal vasculature abnormalities.
Methods: VisionQuest Biomedical developed the CARV tool,
which integrates fully automatic detection algorithms for vessel
abnormalities in order to assess the presence of retinal features
associated to cardiovascular diseases. CARV is composed of two
modules: a) Vessel network analysis, and b) Vessel abnormality
detection.
For the vessel network analysis, the vasculature of the retinal images
is segmented and the arteries and veins are automatically classified
using color information and morphological features. These processes
remove a significant time burden from readers using a 2nd reader
system and enable a fully automatic 1st reader system.
The vessel abnormalities detected and measured are: tortuosity, the
presence of artery-vein (AV) nicking at AV crossings, the AV ratio,
arterial copper/silver wiring, and the presence of retinal emboli in
arteries vessel sections. These algorithms are integrated in a graphical
user interface (GUI) to assist the reader in the detection of vessel
abnormalities and disease.
We evaluated the impact of CARV by enrolling two newly certified
readers and one optometrist to read the images without CARV and to
perform a second read using CARV after a period of memory erase. A
total of 120 fundus images (30% were CVD cases) were selected for
this analysis and were given to each reader.
Results: We compared the sensitivity of the readers against the
ground truth for each image as well as the reading time with and
without CARV. Results are shown in Table 1. The software-aid
resulted in a statistically significant reduction of reading time per
image (30% less). Usage of CARV also improved the sensitivity by
an average of 16%. We measured the inter-reader agreement using
Gwet’s AC1 and observed improvement for the disease detection
from 0.29 to 0.70.
Conclusions: Software-assistance using CARV demonstrated
significant improvement in reader’s performance in terms of
accuracy, consistency, and reading efficiency. This system can
benefit tele-retinal screening for increasing grading throughput and
consistency, and could also be used as a learning tool for retinal
readers.
results. Speed, ease of use, and clinical effectiveness were the main
criteria in system development.
Table 1: Effect of CARV on reading time and sensitivity.
Commercial Relationships: Carla Agurto Rios, Visionquest
Biomedical LLC (E); E Simon Barriga, Visionquest Biomedical
LLC (E); Sheila C. Nemeth, Visionquest Biomedical LLC (E);
Elizabeth McGrew, Visionquest Biomedical LLC (E); Cesar
Carranza, Visionquest Biomedical LLC (E); Dalia C. Guadarrama
Vallejo, None; ETHEL B. GUINTO ARCOS, None; Peter Soliz,
Visionquest Biomedical LLC (I); Vinayak S. Joshi, Visionquest
Biomedical LLC (E)
Support: NEI grant 1R43EY024169-01
ReVMS (Retinal Vasculature Measurement System)
Jeffrey C. Wigdahl, Pedro Guimaraes, Enea Poletti, Alfredo Ruggeri.
Information Engineering, University of Padova, Padova, Italy.
Purpose: This work provides ophthalmologists with a semiautomatic tool for the complete analysis of the retinal vasculature.
It will allow clinicians and researchers a quantitative assessment of
clinical vessel parameters, provide a framework to test new automatic
algorithms, and have the ability to edit results at several steps in
process.
Methods: The application combines automatic algorithms for vessel
segmentation, optic disc detection, and Artery/Vein classification,
with the ability to edit vessel segments, for the calculation of
common clinical vessel parameters (tortuosity, artery/vein ratio
(AVR), fractal dimension, AV nicking). After vessel segmentation,
the user can edit vessel segments (add new, remove, combine,
split). Vessel widths, tortuosity metrics, and fractal dimension
are calculated. Optic disc center and radius are found and can be
manually adjusted.
Artery/Vein classification is performed in an area between 1 and 2
disc diameters from the optic disc. The user has the ability to edit,
add/remove vessels, and change vessel widths before automatic
AV classification. The user can edit results again before AV ratio,
central retinal vein equivalent, and central retinal artery equivalent
calculation. Finally, vessel crossover points are analyzed to detect
narrowing and angle in/out. Vessel widths can be edited prior
to calculation. All clinical parameters are computed using our
previously developed methods. Tortuosity is measured per segment
and combined to give a whole image tortuosity. The set of 45
images used to test the system are from the High Resolution Fundus
(HRF) image database provided by Friedrich-Alexander University
Erlangen-Nuremberg. Images are 2336x3504 pixels covering a 50
degree field of view.
Results: Intermediate steps of the system can be seen in Fig. 1. The
system has been used to detect vessels, compute tortuosity and AV
ratio in a set of 45 high quality images.
Conclusions: This system provides a framework for the quantitative
analysis of the retinal vasculature with the intention of making the
automatic portions interchangeable, for the testing of new algorithms,
while maintaining editing functions, to ensure the highest quality
Fig.1 Top Left- Original Image. Top Right- Binary vessel image.
Bottom Left- Visualization of Twist-based tortuosity metric. Bottom
Right- AVR interface with classified vessels.
Commercial Relationships: Jeffrey C. Wigdahl, None; Pedro
Guimaraes, None; Enea Poletti, None; Alfredo Ruggeri, None
Support: Made possible by a Marie Curie grant from the European
Commission in the framework of the REVAMMAD ITN (Initial
Training Research network), Project number 316990.
Spectral Classification of Retinal Features Using K-Means
Clustering Algorithm
Julie Cho1, Amir H. Kashani1, 2, Mark S. Humayun1, 2. 1Keck School
of Medicine of USC, Los Angeles, CA; 2Ophthalmology, USC Eye
Institute, Los Angeles, CA.
Purpose: To develop effective methods of sampling and classifying
retinal tissue spectra obtained with a hyperspectral computed
tomographic imaging spectrometer.
Methods: High-resolution hyperspectral images (Figure 1a) from
three species (rabbits, canine and minipig) were acquired using a
hyperspectral computed tomographic imaging spectrometer (HCTIS)
that has been previously described in detail (Kashani AH et al., 2011,
2012, 2013). Animals were sedated and imaging was performed
under general anesthesia through pharmacologically dilated pupils
using an HCTIS mounted on a Zeiss FF450 fundus camera. The
HCTIS records up to 76 spectral bands from 450-700nm within a
single snapshot obtained by a standard fundus photograph. For this
study, multiple images of characteristic regions from each animal
were obtained including arteries, veins, retina and optic disc. Spectra
from multiple regions of interest within individual images, across
images, and across species were manually inspected for similar
spectral characteristics. Spectra were also quantitatively analyzed
with k-means clustering algorithm and Voronoi diagram was
generated (Figure 1d).
Results: Spectral characteristics of retinal vessels and tissue are
robust and largely correlated to vascularity (Figure 1b-d) and
pigmentation in the animal models. Vascular structures were
reliably identified across multiple species and required minimal
dimensionality. Three-dimensional data were best clustered with
differing k-values (k=3 dog, k=2 pig, k=1 rabbit). Features such
as retinal pigmentation across species could be determined by
characteristic heterogeneity of each tissue as determined by standard
deviation of the mean (Figure 1e). Characterization of these structures
required multi-dimensional models, but segmentation was possible
with silhouette means above 0.6 in the case k=1 and above 0.8 in case
k=2 and k=3 indicating compactness of the data (Figure 1e).
Conclusions: Characterization of retinal features in multiple spectral
dimensions allows robust and reliable classification of retinal tissue.
Here we sampled multiple images and found defining spectral
characteristics of structural features within and across images as well
as across species. These results suggest that clustering methods can
be used to help classify tissue types in hyperspectral images.
repeated 20 times to test all images. In our experiments n=60 and
final classification is obtained using majority vote. The classified
image is further refined by applying post processing to remove
outliers.
Results: The proposed classifier is 88.98% sensitive and 91.05%
specific for vessel classification in complicated images before post
processing and it is improved to 90.10% and 90.13% respectively
after post processing with the area under the curve (ROC) of 0.9006.
The example of STARE image, ground truth and segmented vessels
by our method is shown in Fig.2.
Conclusions: The proposed method classifies pixels of 2D fundus
images as vessel or non-vessel. It uses patch based PCA and LDA to
extract features and classifies them using ensembles of decision trees.
The results show vessels are segmented effectively in normal and
complicated images.
Figure1. Spectral Analysis of the Retina.
Commercial Relationships: Julie Cho, Reichert Technologies (F);
Amir H. Kashani, Reichert Technologies (F); Mark S. Humayun,
Reichert Technologies (F)
Vessels segmentation in color retinal images using ensemble of
bagged decision trees and patched based principle component
analysis and linear discriminant analysis
Ehsan V. Shahrian, Siamak Yousefi, Arash Mozayan Isfahani,
Hema L. Ramkumar, Joseph T. Nezgoda, Christopher Bowd, Felipe
A. Medeiros, Linda M. Zangwill, Robert N. Weinreb, Michael H.
Goldbaum. Ophthalmology, University of California, San Diego, San
Diego, CA.
Purpose: To describe and test a novel, robust method for automatic
segmentation of vessels in complicated color retinal images with
confounding vessel-like objects.
Methods: This method classifies pixels of 2D fundus images
as vessel or non-vessel pixels. Twenty 35 degree field of view
photographic retinal images from the STARE database collected at
the UC San Diego, manually annotated by an expert (MHG), were
used as a training and test set for the classifier. An example STARE
image and its annotated ground truth vessels are shown in Fig1.a.
Intensities of pixels in the square neighborhood of pixels were used
to build raw feature vectors. The square neighborhood of size 9x9
pixels was used to build an 81-dimensional raw feature set for every
pixel. The square neighborhood for pixel p is highlighted with yellow
window in Fig1.a. The 81-dimensional feature set was reduced to 7
dimensions through a two-stage dimensional reduction process as
shown in Fig1.b. In the first stage, principal components analysis
(PCA) was applied to reduce dimensions of feature set and then
linear discriminant analysis (LDA) was employed in the second stage
to map vessel and non-vessel pixels into new spaces that are more
discriminant. The classifier is constructed based on ensembles of n
decision trees as shown in Fig1.c. To provide independent training
and test image sets, leave-one-out methodology was used to train
classifiers on 19 images and to test on one image with the process
Support: P30EY022589 and participant retention incentive grants in
the form of glaucoma medication at no cost from Alcon Laboratories
Inc, Allergan, Pfizer Inc, and Santen Inc.Unrestricted grant from
Research to Prevent Blindness, New York, New York, EY11008,
U10EY14267, EY019869, EY021818, EY022039, P30EY022589
Eyesight Foundation of Alabama; Alcon Laboratories Inc.; Allergan
Inc.; Pfizer Inc.; Merck Inc.; Santen Inc.; and the Edith C. Blum
Research Fund of the New York Glaucoma Research Institute, New
York, NY, Unrestricted grant from Research to Prevent Blindness,
New York, New York
Commercial Relationships: Ehsan V. Shahrian, None; Siamak
Yousefi, None; Arash Mozayan Isfahani, None; Hema L.
Ramkumar, None; Joseph T. Nezgoda, None; Christopher
Bowd, None; Felipe A. Medeiros, Carl Zeiss Meditec Inc (F),
Alcon (F), Alcon Laboratories (R), Allergan (C), Allergan Inc (F),
Bausch & Lomb (F), Carl Zeiss Meditec (R), Carl-Zeiss Meditec
(C), Heidelberg Engineering (F), Merck Inc (F), National Eye
Institute (F), Novartis (C), Reichert (F), Reichert Inc (R), Sensimed
(F), Topcon (F); Linda M. Zangwill, Carl Zeiss Meditec Inc (F),
Heidelberg Engineering GmbH (F), Nidek (F), Optovue Inc (F),
Topcon Medical Systems Inc (F); Robert N. Weinreb, Aerie (F),
Alcon (F), Allergan (F), Amatek (F), Aquesys (F), Bausch&Lomb
(F), Carl Zeiss Meditec (F), Carl Zeiss Meditec (F), Genentech (F),
Heidelberg Engineering GmbH (F), Nidek (F), Novartis (F), Optovue
(F), Topcon (F), Topcon (F), Valeant (F); Michael H. Goldbaum,
None
Automated vasculature segmentation in retinal images using
multi-scale image analysis
Malavika Bhaskaranand, Chaithanya Ramachandra, Sandeep Bhat,
Kaushal Solanki. Eyenuk, Inc., Woodland Hills, CA.
Purpose: Abnormalities in retinal vasculature provide useful
information about clinical and sub-clinical cerebrovascular,
cardiovascular, and metabolic health of the patient. This is especially
true for diabetic retinopathy (DR), a common microvascular
complication of diabetes in which damaged retinal blood vessels
become leaky or occluded, ultimately leading to visual loss. Despite
its promise, current clinical practice does not include retinal vessel
analysis for DR primarily because the analysis involves highly
laborious task of extracting vessels. We present a segmentation
method that can automatically mark retinal vasculature and thus aid
in automated DR screening.
Methods: Our vasculature segmentation method utilizes novel and
customized image analysis techniques including image normalization,
multi-scale putative vessel detection, and multi-orientation
morphological filterbank analysis for vessel estimation.
We evaluate our segmentation method on the publicly available
DRIVE and STARE retinal image datasets with vasculature marked
in great detail. We compute the accuracy and the area under the
receiver operating characteristic (AUROC) at pixel-level for our
segmentation method using all the images in each dataset.
Results: Vessel segmentation maps generated by our vasculature
segmentation method for three retinal images are shown in Figure
1. Our method is robust to the varying pixel intensity ranges and
qualities of the retinal images. Our method achieves accuracy of
95.3% and AUROC of 0.932 on the DRIVE dataset and accuracy of
95.6% and AUROC of 0.914 on the STARE dataset. These results are
equivalent to the performance of a human grader.
Conclusions: We present a new approach for retinal vasculature
segmentation that has the potential for real-world use with its high
accuracy, invariance to imaging conditions, and good generalizability.
on a publicly available dataset and the comparison with our previous
method demonstrate its better performance.
Fig1: Example vessel segmentation maps. Top row – Original retinal
fundus images. Bottom row – corresponding automated vessel
segmentation maps.
Commercial Relationships: Malavika Bhaskaranand, Eyenuk,
Inc. (E); Chaithanya Ramachandra, Eyenuk, Inc. (E), Eyenuk, Inc.
(P); Sandeep Bhat, Eyenuk, Inc. (E), Eyenuk, Inc. (P); Kaushal
Solanki, Eyenuk, Inc. (E), Eyenuk, Inc. (P)
Support: NIH STTR Phase II Grant 2R44EB013585-03
Automated Construction of Arterial and Venous Trees on Retinal
Images Using Topological and Intensity Information
Qiao Hu1, Mona K. Garvin1, 3, Michael D. Abramoff2, 1. 1Electrical and
Computer Engineering, Univ of Iowa, Iowa City, IA; 2Ophthalmology
& Visual Sciences, The University of Iowa Hospitals & Clinics, Iowa
City, IA; 3Center for the Prevention and Treatment of Visual Loss,
Veterans Administration Hospital, Iowa City, IA.
Purpose: Automated retinal vessel analysis from fundus images,
including central retinal artery and venous equivalent estimations,
requires automated labelling of arteries and veins. We previously
developed a graph-based framework to separate overlapping arterialvenous (A/V) trees (Hu et al., MICCAI 2013). Here we present a new
version to determine arteries and veins in overlapping as well as nonoverlapping vessels by including a pixel classification algorithm.
Methods: An expert annotated vessel pixels in the public dataset
DRIVE (40 images from 40 subjects, equally divided into a training
and a test set) as artery or vein (Fig. 1). Our approach first uses the
topology of the vasculature to separate overlapping vessels into
A/V trees with a graph-based algorithm. Then a support-vectormachine classifier (trained on the training set using 19 local intensity
features) is used to classify independent vessels into A/V vessels. The
approach is validated on the test set, with both manual and automatic
vesselness maps as inputs. The coverage rate (ratio of classified
vessel pixels over all vessel pixels defined in the A/V tree reference
standard) and the accuracy (ratio of correctly classified vessel pixels
over all classified vessel pixels) are used in the evaluation.
Results: An example result is shown in Fig. 2. The mean accuracy/
coverage rate with 95% confidence interval was 88.2% (84.3, 92.1)
/ 88.5% (86.8, 90.2) for the manual segmentation; and 82.5% (77.4,
87.6) / 85.19% (83.7, 86.7) for the automatic segmentation. Using
our previously developed approach, the mean accuracy/coverage rate
with 95% CI was 89.06% (84.9, 93.3) / 82.03 (80.0, 84.1) for the
manual segmentation; and 83.08% (77.6, 88.5) / 78.6% (77.0, 80.2)
for the automatic segmentation. Thus the approach improves the
coverage rate significantly with similar accuracy for both inputs.
Conclusions: Here we present a method to automatically construct
the A/V trees in retinal images given a vessel segmentation. The test
Fig. 1 (a) A fundus image from DRIVE (b) Reference A/V trees
(red=artery, blue=vein, green=overlapping, white=uncertain)
two independent clinicians. Intra- and inter-grader agreement
was measured. Composite annotations were created for algorithm
validation. An automated algorithm for detection of abnormal
capillaries in BRVO was prospectively tested against these composite
annotations.
Results: Near-perfect intra-grader agreement (Fleiss’ kappa range
0.81 - 0.92) and substantial inter-grader agreement (Cohen’s kappa
range 0.73 – 0.84) was found among human annotations. Substantial
agreement was found between composite annotations and the output
of the fully automated detector (Cohen’s kappa for Case 1: 0.59, Case
2: 0.72, Case 3: 0.58, Case 4: 0.69).
Conclusions: Our fully automated algorithm showed good agreement
with human expert annotations.
Commercial Relationships: David C. Reed, None; Sanket U.
Shah, None; Kris Zutis, None; Emanuele Trucco, None; JeanPierre Hubschman, None
Fig. 2 (a) Vesselness maps (b) Results using prior method (c) Results
using proposed method
Commercial Relationships: Qiao Hu, No. PCT/US2014/028055
(P); Mona K. Garvin, No. PCT/US2014/028055 (P); Michael D.
Abramoff, IDx LLC (I), No. PCT/US2014/028055 (P)
Support: NIH R01 EY018853;the Department of Veterans Affairs
Rehabilitation Research and Development Division (I01 CX000119
and Career Development Award IK2RX000728);the Marlene S. and
Leonard A. Hadley Glaucoma Research Fund
Fully Automated Detection of Abnormal Capillaries in UltraWide-Field Fluorescein Angiography of Branch Retinal Vein
Occlusion
David C. Reed1, 2, Sanket U. Shah3, 2, Kris Zutis4, Emanuele Trucco4,
Jean-Pierre Hubschman2. 1Retina, Wills Eye Hospital, Philadelphia,
PA; 2Jules Stein Eye Institute, University of California Los Angeles,
Los Angeles, CA; 3Eugene and Marilyn Glick Eye Institute, Indiana
University, Indianapolis, IN; 4School of Computing, University of
Dundee, Nethergate, United Kingdom.
Purpose: Our objective was to develop a prototype fully automated
detector for capillary abnormalities in ultra-wide-field fluorescein
angiograms (UWFA) of branch retinal vein occlusion (BRVO).
Methods: Four new UWFAs of BRVO were obtained. For each
case, three annotations of abnormal capillaries were created by
Automated Detection of Microaneurysms Using Curvelet
Transform
Syed Ayaz Ali Shah1, Tong Boon Tang1, Augustinus Laude2, Ibrahima
Faye3. 1Department of Electrical and Electronic Engineering,
Universiti Teknologi PETRONAS, Tronoh, Malaysia; 2Tan Tock
Seng Hospital, National Healthcare Group Eye Institute, Singapore;
3
Department of Fundamental and Applied Science, Universiti
Teknologi PETRONAS, Tronoh, Malaysia.
Purpose: Microaneurysms (MAs) are said to be the first sign
of diabetic retinopathy (DR). Counting the number of MAs is
important for monitoring the progression of DR. Factors making
MA detection a challenge include the variation in MA size, low and
varying image contrast, uneven illumination and variation in fundus
image background. Our objective is to design a fully automatic MA
detection system with a high sensitivity and specificity.
Methods: MAs and blood vessels were selected by using local
thresholding technique in green channel. Artefacts due to image
background were identified by contrast, variance and standard
deviation information; blood vessels were segmented by wavelet
transform technique. Removal of background and blood vessels
from the fundus images left initial MA candidates. Based on shape
parameters (eccentricity, aspect ratio and circulatory) and calculated
features (energy, standard deviation and variance), the candidates
were filtered using Curvelet coefficients. To further reduce false
positives, color features like intensity in HSV channels and standard
deviation and variance in HSV and RGB channels were used.
Results: Validation was performed using publically available
‘Retinopathy Online Challenge (ROC)’ dataset. The dataset consists
of 50 training images with ground truth and 50 test images. Out of
those 50 training images, 37 images contain MAs (336 MAs in total).
The proposed algorithm was evaluated with all those 37 images. Our
method detected 125 MAs thus achieving 37.2 % sensitivity with
74.89 false positives per image (FPPI). The experiment result was
compared with published results (refer to Table 1). Adal achieved
better sensitivity and specificity by using a total of 64-SURF
descriptors, three features from Radon space and two image patch
features, in contrast to our 19 features (6 color features, 5 shape
parameters and 8 statistical features). Lazar constructed peak map
and extracted MA candidates using hysteresis thresholding. Regions
of interest (ROIs) were manually selected. Our approach is however
fully-automated.
Conclusions: A MA detection system based on local thresholding
and Curvelet transform has been proposed. The system performance
was evaluated with other existing techniques. Compared with the
state-of-the-arts, it uses significantly less feature size (hence faster
computation) and is fully automated while achieving comparable
results.
Table 1. Comparison of different published methods.
Commercial Relationships: Syed Ayaz Ali Shah, None; Tong Boon
Tang, None; Augustinus Laude, None; Ibrahima Faye, None
Support: NHG CSCS/120006
Masking Vasculature and Measuring Fundus Autofluorescence
using Standard Grids
Kenneth R. Sloan1, 2, Fazila Aseem2, Anna V. Zarubina2, Mark E.
Clark2, Cynthia Owsley2, Christine A. Curcio2. 1Computer and
Information Science, UAB, Birmingham, AL; 2Department of
Ophthalmology, UAB, Birmingham, AL.
Purpose: Create a semi-automatic workflow to establish a frame of
reference for standard grids (e.g., ETDRS), identify regions for AF
analysis, and visualize results.
Methods: A custom FIJI plug-in segmented Spectralis AF images
and guided a trained observer in editing a mask excluding vessels.
We selected the Phansalkar adaptive local thresholding method (a
standard Fiji tool) applied over circular regions of r= 200 pixels.
Images were registered using the foveal center and scale information
from the instrument. Stand-alone Java programs accepted the original
AF image, the mask, and scale and location information and tabulated
AF intensities for unmasked pixels grouped by regions within
standard grids. Statistical properties including texture measures were
computed for grid regions and presented as tables and as custom
visualizations.
Results: A total of 660 images (1536x1536 8-bit grayscale) from
the Alabama Study on Age-Related Macular Degeneration were
processed. The Phansalkar method produced a satisfactory initial
segmentation in a few seconds per image. Manual editing required
10-30 minutes per image, depending on image complexity and
experience of the editor. The final segmentations were judged to be
very good to excellent.
Conclusions: Standard Fiji tools can segment retinal vasculature and
other excluded features when augmented by a final manual editing.
Masking the vasculature and other features is superior to histogrambased methods, and extends AF measurement in retinal images to a
broader field of view.
Left: AF image of macula with grid overlay; sampled pixels are
green, masked pixels are red.
Middle: display of mean (top), coefficient of variation (middle), and a
composite using hue and intensity to show both.
Right: histograms of sampled pixel intensities in each region.
Commercial Relationships: Kenneth R. Sloan, None; Fazila
Aseem, None; Anna V. Zarubina, None; Mark E. Clark, None;
Cynthia Owsley, None; Christine A. Curcio, None
Support: 5R01AG004212, EY06109; Research to Prevent Blindness,
Inc.; EyeSight Foundation of Alabama.
Sampling paradigms for increasing the repeatability of Static
Vessel Analysis by using multiple images
Steffen Rieger1, Walthard Vilser2, Daniel Baumgarten1. 1BMTI, TU
Ilmenau, Ilmenau, Germany; 2Imedos Systems UG, Jena, Germany.
Purpose: Static Vessel Analysis is an accepted tool for risk
evaluation of cardiac diseases based on retinal vessel diameters
analyzed in fundus images. One reason for uncertainties between
repetitions is the biological temporal variation of the vessels.
Standard deviation (SD) between repetitions can be reduced by
analyzing multiple images per measurement. We developed sampling
paradigms using additional information about vessel activities and
investigated their advantage in contrast to randomized acquired
samples.
Methods: Main signal components in the temporal variation of
retinal vessel diameters were determined from literature. With respect
to the Mayer waves as main components of the biological variation of
the vessel diameters, paradigms were developed to sample the waves
with phase shifts of 180 degree. For the first evaluation, a dataset of
signals containing the main frequencies was simulated, considering
the individual variation of these frequencies. The signal was sampled
several times according to the developed paradigms and the mean
value was calculated. SD between multiple measurements was
determined and compared to SD of measurements with randomized
sampling times. The influence of the precision of knowing the
signal frequencies was observed. Following this, continuous vessel
diameters were analyzed, recorded by Dynamic Vessel Analyzer
under mydriatic conditions. The sampling paradigms were applied to
these real vessel diameter data.
Results: The acquisition and evaluation of multiple images improves
the reproducibility of the determination of the vessel diameters.
Our simulations reveal that SD between repetitions is reduced by
1/√n when randomly sampling n images, whereas sampling with
the developed paradigms further reduces SD, in particular for small
numbers of images. The benefit is high even with variations of the
main signal frequencies. With an increasing number of samples,
improvement decreases again. These results could be verified on
measurement data.
Conclusions: According to our investigations, uncertainties
induced by temporal biological variations can be highly reduced by
the acquisition of small numbers of images taken with dedicated
sampling paradigms. By recording and evaluating of a small image
series repeatability can be considerably increased. These images can
be taken with a fundus camera under non-mydriatic conditions with a
low strain on the patients.
Commercial Relationships: Steffen Rieger, None; Walthard
Vilser, None; Daniel Baumgarten, None
Support: German Federal Ministry of Education and Research,
Grant 3IPT605X
Diameters of Large Retinal Blood Vessels in Hypertensive
Patients as Measured by Spectral Domain Optical Coherence
Tomography
Amit Meshi1, Jonathan Shahar2, Yaron Arbel3, Shlomo Berliner3, Anat
Loewenstein2, Dafna Goldenberg2. 1Ophthalmology, Meir Medical
Center, Kfar Sava, Israel; 2Ophthalmology, Tel Aviv Medical Center,
Tel Aviv, Israel; 3Internal Medicine D+E, Tel Aviv Medical Center,
Tel Aviv, Israel.
Purpose: To measure retinal blood vessel diameters in hypertensive
patients using spectral domain optical coherence tomography (SDOCT).
Methods: A cohort of 47 hypertensive patients (94 eyes) underwent
an SD-OCT exam (Spectralis, Heidelberg Engineering, Heidelberg,
Germany). Two cubes of seven high-resolution horizontal scans each
were placed at the superior and inferior borders of the disc to include
the large temporal retinal vessels (Figure 1). Inter-scan interval was
240 mm. The outer diameter of the main temporal arteriole and venule
was measured manually (Image J software, National Institute of
Health, USA) at distances 480 mm, 720 mm, 960 mm, 1200 mm and
1440 mm from the optic disc border, superiorly and inferiorly (Figure
2). Previously reported results from 29 healthy subjects (58 eyes)
were used as controls.
Results: The mean ± SD age of the hypertensive patients was
57.3 ± 8.85 years. Their mean ± SD systolic and diastolic blood
pressures were 140.23 ± 13.22 mmHg and 82.17 ± 10.68 mmHg,
respectively. The mean ± SD diameters of the arterioles and venules
in the hypertensive cohort steadily decreased from 130.85 ± 13.7
mm and 162.02 ± 16.04 mm at 480 mm from the optic disc to 119.7
± 13.45 mm and 151.35 ± 15.44 mm at 1440 mm from the optic
disc, respectively. The overall mean artery-to-vein ratio (AVR)
in hypertension was 0.81. Uncontrolled hypertensive patients
had narrower mean arteriolar diameters compared to controlled
hypertensive patients at all points of measurement, reaching statistical
significance in the superior arterioles (P = 0.009). Compared to
healthy controls, mean arteriolar diameters and mean AVR were
smaller (P < 0.04), whereas mean venular diameters were larger (P <
0.01) at all points of measurement in hypertensive patients.
Conclusions: Measurement of large temporal retinal vessel diameters
by SD-OCT may be used as an adjunctive tool for the assessment of
hypertension.
Inferior cube of the left eye demonstrating the 7 scans with a 240 mm
inter-scan interval. Each arrow indicates the distance from the optic
nerve edge.
Measurement of the retinal venule (V) and arteriole (A) diameters.
The black and red arrows indicate the borders of the venule and
arteriole, respectively, where the hyper-reflective signal is seen. The
white arrow represents the caliper used for retinal vessel diameter
measurement.
Commercial Relationships: Amit Meshi, None; Jonathan
Shahar, None; Yaron Arbel, None; Shlomo Berliner, None; Anat
Loewenstein, None; Dafna Goldenberg, None
Total Retinal Blood Flow Measurement of Normal and Diabetic
Eyes with 100k Hz Swept Source Domain OCT
Ou Tan, Liang Liu, Gangjun Liu, Yali Jia, Andreas K. Lauer, David
Huang. Casey Eye Institute, Oregon Health & Science Univ,
Portland, OR.
Purpose: To evaluate total retinal blood flow (TRBF) using 100k Hz
Doppler swept-source OCT for healthy and diabetic eyes.
Methods: One eye of each participant was scanned with a 100k
Hz prototype swept source OCT. Eye length was measured with
IOL master 500 (Carl Zeiss, CA). The TRBF scan pattern contains
8 repeated volume scans (depth=2.3mm, horizontal=1.6 mm, and
vertical=2mm) obtained in 4 seconds, and centered on central retinal
vessels. Each eye was scanned 3 times. Phase noise was reduced
using a novel algorithm that removes timing jitter from the tunable
laser k-clock using information from fixed pattern artifacts. Tissue
bulk-motion was removed using a histogram method. An automated
algorithm was developed for TRBF measurement using multiple
plane en face technique. For each retinal vein, blood flow was
measured at an optimal en face plane where the calculated flow is
maximized. The TRBF was calculated by summing flow in all veins.
The algorithm tracks vascular branching so that either root or branch
veins are summed, but never both. The TRBF in 8 repeated volumes
were averaged to reduce variation due to pulsation during the cardiac
cycle. Finally the TRBF is corrected by multiplying with square of
the ratio between eye length and default eye length.
Results: A total of 12 eyes of 12 participants (6 healthy controls, 6
diabetic participants) were enrolled. The TRBF of healthy controls
is 46.4±5.4 ml/min. The TRBF of diabetic eyes is 36.4±12.8 ml/min.
The intra-visit repeatability was 6.1% for health control and 8.4% for
diabetic eyes.
Conclusions: An automated algorithm was developed for TRBF
measurement using 100 kHz spectral-domain OCT. Diabetic eyes
had significant lower TRBF than normal subject. The repeatability of
TRBF was improved compared with previous methods and this may
be useful for the monitoring of glaucoma and diabetic retinopathy.
Commercial Relationships: Ou Tan, Optovue (F), Optovue (P),
Zeiss meditec (P); Liang Liu, None; Gangjun Liu, None; Yali Jia,
Optovue (F), Optovue (P); Andreas K. Lauer, None; David Huang,
Optovue (F), Optovue (I), Optovue (P), Zeiss Meditec (P)
Support: NIH grants R01 EY023285, DP3 DK104397, CTSA grant
(UL1TR000128), NIH and an unrestricted grant from Research to
Prevent Blindness to Casey Eye Institute.
Comparison of the manually delineated Foveal Avascular Zone
on Traditional Fluorescein Angiography and Doppler Optical
Coherence Tomography Images acquired using a 1060nm SweptSource Speckle-Variance Doppler-OCT Angiography Prototype
Christoph Mitsch1, Jan Lammer1, Laurin Ginner2, Daniel Fechtig2,
Cedric Blatter2, Rainer A. Leitgeb2, Ursula Schmidt-Erfurth1.
1
Ophthalmology and Optometrics, Medical University of Vienna,
Vienna, Austria; 2Department of Medical Physics and Biomedical
Engineering, Medical University of Vienna, Vienna, Austria.
Purpose: Doppler-optical coherence angiography (DOCT-A)
represents a non-invasive alternative to traditional, dye-based
fluorescein angiography (FA). Due to the considerable difference of
the image acquisition and computation techniques of both modalities,
the comparability of the results needs to be evaluated.
Methods: On a single-day visit, patients underwent both FA and
DOCT-A examinations. Early phase FA images were exported and
en-face projections of of the DOCT-A image stacks were generated.
The resulting dataset was duplicated and masked. Two graders
delineated the foveal avascular zone (FAZ) on every image of each
patient and modality two times. The resulting FAZ area values were
compared by calculating the Spearman Product-Moment Correlation
Coefficient to illustrate inter- and intra-grader reproducibility in both
modalities and the intra-modality reproducibility.
Results: 22 eyes of 18 patients with different retinal pathologies
were imaged. The FAZ areas (in square degrees) delineated in
DOCT-A images measured 2.97±1.98 (range 0.68-9.24) in grader
1 and 3.69±2.62 (range 0.99-11.89) in grader 2. On FA, the areas
measured 3.51±2.32 (range 0.83-10.33) in grader 1 and 5.15±3.89
(range 1.58-17.47) in grader 2, respectively. The intra-grader FAZ
delineation correlation coefficients in DOCT-A and FA were 0.88 and
0.99 in grader 1 and 0.8 and 0.83 in grader 2. The resulting intergrader DOCT-A and FA correlation coefficients were 0.75 and -0.25,
respectively. Inter-modality correlation coefficients were 0.36 in
grader 1, 0.1 in grader 2 and 0.29 for the means.
Conclusions: The FAZ area delineated on DOCT-A images showed
strong inter-grader correlation, higher than in FA images, which was
only weak. Inter-modality correlation was weak in one grader and
in the grader means and very weak in the other grader. Although
these results may reveal a wide range of interpretability differences
for both modalities, the strong inter-grader correlation in DOCT-A
FAZ delineation detected in our data is promising. The weak intramodality comparability needs to be interpreted accounting for the
elemental differences of both image acquisition techniques and their
resulting individual sources of quality- and interpretability cutbacks.
Commercial Relationships: Christoph Mitsch, None; Jan
Lammer, None; Laurin Ginner, None; Daniel Fechtig, None;
Cedric Blatter, None; Rainer A. Leitgeb, None; Ursula SchmidtErfurth, None
A Sloped Piece Gaussian Model for Foveal Pit Shape Tested on
OCT Data from Young Adults
Lei Liu1, Wendy Marsh-Tootle1, Elise N. Harb5, Leslie Hyman2, Wei
Hou2, Qinghau Zhang2, Heather A. Anderson4, Thomas T. Norton1,
Jane E. Gwiazda3. 1School of Optometry, Univ of Alabama at
Birmingham, Birmingham, AL; 2Preventive Medicine, Stony Brook
Medicine, Stony Brook, NY; 3New England College of Optometry,
Boston, MA; 4College of Optometry, University of Houston,
Houston, TX; 5School of Optometry, University of California at
Berkeley, Berkeley, CA.
Purpose: Interest in foveal pit shape has grown due to its possible
links to race, gender, refraction, amblyopia and retinal pathology
and the increasing availability of high-quality OCT scans. Foveal
pits with flat bottoms and/or asymmetric walls were poorly fitted
using the Difference of Gaussians (DoG) models in previous studies.
We explored more sophisticated mathematical models to improve
goodness of fit and to capture important anatomic characteristics.
Methods: Raw data from 3501 EMM5 foveal scans (RTVue 100)
from 582 young adults (421 myopes and 161 matched non-myopes,
mean age=21.9±1.4 yrs) were used to test fitting of the DoG model
and a Sloped Piece Gaussian (SPG) model. Three consecutive
scans were obtained from 99.4% of the eyes. The SPG was a linear
combination of a tilted line and a piecemeal Gaussian function (two
halves of a Gaussian connected by a straight line) with 6 parameters
(the σ of the Gaussian, the center (C) and width (P) of a flat bottom,
the slope of the tilted line (T), the height and vertical position of
the curve. See figure). The 803-pixel (6 mm) horizontal and vertical
retinal thickness curves (ILM to RPE) passing through the foveal
center were first smoothed using an 80-pixel span to determine the
first local maxima on both sides of the pit. The SPG model was then
fitted to the pit data between the two maxima. Data with 50 additional
pixels beyond the two maxima were fitted by the DoG. A nonlinear
least squares method was used to find the best fitting curves. SPG and
DoG were compared using t-tests.
Results: The SPG model produced significantly smaller root mean
square errors than the DoG (3.9 vs. 7.1 μm, t=88.8, p<0.0001,
vertical; 4.2 vs. 10.9 μm, t=116.3, p<0.0001, horizontal). The SPG
revealed a consistent tilt in >95% of the curves with the nasal side
higher than the temporal side by 11.5±0.70 μm (t>32.9, p<0.0001),
similar to the 13 μm asymmetry from a histologic study (Curcio
et al, IOVS, 2011). Such asymmetry was much less pronounced in
vertical scans and might be related to the papillomacular bundle. The
pit bottom width ranged from 0 to 600 μm, with 40% and 54% of the
pits having a flat bottom >30 μm in horizontal and vertical directions,
respectively. None of these characteristics was captured by the DoG
model.
Conclusions: A SPG model provided superior fits to OCT
foveal scans and captured a wider range of foveal pit anatomical
characteristics than previous models.
Further, only 3 of the 12 eyes fell outside the 95% confidence limits
(CI) on one or more of the 3 metrics (all 3 for S and 2 for W and
D). The 12 eyes were on average older, but the difference was not
significant (44.8±15.2 vs. 37,8±14.3yrs, p=0.11).
Conclusions: Variations in foveal pit anatomy do not provide a
reliable indicator of circum-foveal FPs seen on OCT inner retinal
thickness maps. These FPs are partially due to normal variations in
RGC+ thickness that are not well correlated with foveal pit anatomy.
Commercial Relationships: Lei Liu, None; Wendy Marsh-Tootle,
None; Elise N. Harb, None; Leslie Hyman, None; Wei Hou, None;
Qinghau Zhang, None; Heather A. Anderson, None; Thomas T.
Norton, None; Jane E. Gwiazda, None
Support: NIH Grants: EY11805, EY11756, EY11754, EY11740,
EY11755 and EY11752
Inter-individual Variations in Foveal Pit Anatomy Do Not
Completely Explain False Positives Seen in the Central Retina on
OCT Inner Retinal Thickness Maps
Khushmit Kaur, Hassan Muhammad, Diane Wang, Jacoby Shelton,
Donald Hood. Columbia University, New York, NY.
Purpose: In healthy controls, apparent circum-foveal abnormalities
can be seen on retinal ganglion cell plus inner plexiform (RGC+)
thickness maps derived from cube scans obtained with optical
coherence tomography (OCT). To better understand these false
positives (FPs), the variation in foveal pit morphology was studied.
Methods: OCT macular cube scans, 6x6 mm, were obtained on one
eye of 126 healthy individuals. The center of the fovea was marked
manually to identify the B-scan corresponding to the horizontal
meridian. A fourth-order negative Gaussian was fitted to the border
between the inner limiting membrane and vitreous. Based upon the 1st
derivative of this Gaussian, the locations of the foveal center and the
perifoveal peaks were taken as the locations of the 2 local maxima
and one local minimum. The horizontal distance from peak to peak
was taken as the foveal width (W), and the average vertical distances
from the location of the fovea to each peak as foveal depth (D). The
steepest slope of each foveal edge was obtained and averaged for a
measure of slope (S). RGC+ thickness was measured from the OCT
cube scan for each eye and probability plots generated (Fig.1) relative
to the other 125 eyes. The circum-foveal region (± 4°, red circle, Fig.
1) was considered abnormal if there were regions falling below the
1% confidence limit in both hemi-retinas.
Results: 12 (9.5%) of the 126 controls met our criteria for circumfoveal abnormalities; 2 are shown in Fig. 1. On average, the fovea
of the 12 eyes was wider, less deep and less steep. However, only
the S values were significantly different than the other eyes at
p≤0.05: W:(222.5±42.1 vs. 212.7±31.9um, p=0.33); D:(38.0±8.2.
vs. 41.4±5.4um, p=0.06); and S:(0.35±0.09 vs. 0.40±0.06; p<0.02).
Fig. 1: The eye in the lower, but not the upper, panel had foveal
measures outside 95% CI.
Commercial Relationships: Khushmit Kaur, None; Hassan
Muhammad, None; Diane Wang, None; Jacoby Shelton, None;
Donald Hood, Topcon, Inc. (F)
Support: R01-EY002115
Asymmetry in Foveal Pit Dimensions
Nancy J. Coletta, Rhett Stroupe. Vision Science, New England
College of Optometry, Boston, MA.
Purpose: Myopic eyes generally have shallower foveal pits than
emmetropic eyes, and we have reported that the width of the foveal
pit decreases with increasing axial length or increasing myopia
(Stroupe and Coletta, ARVO, 2013). The foveal pit exhibits an
asymmetry, in that it is wider in the horizontal than in the vertical
dimension. The purpose of this study was to examine how the
asymmetry in foveal pit dimension varies with axial length.
Methods: Axial lengths and macular thickness were obtained on
both eyes of 25 subjects (average age 24.8 years). Refractions ranged
from plano to -9.90 D. Axial lengths were measured with a Zeiss
IOL Master and retinal thickness was measured with an Optovue
RTVue spectral domain OCT. We used the MM6 scan that consists of
twelve 6-mm length B-scans that rotate through the fovea at 15 deg
intervals. OCT results were analyzed using a custom Matlab program
that computed these parameters for each scan orientation: maximum
thickness on either side of the foveal pit, minimum thickness, depth
of the foveal pit, pit width across the maximum thickness (FWFM)
and pit width at half the pit depth (FWHM). Lateral dimensions
were corrected for the variation in transverse magnification of the
OCT scan against axial length. Results from the vertical scan and
its two adjacent scan orientations were averaged to obtain vertical
dimensions and results from the horizontal scan and its two adjacent
scan orientations were averaged to obtain horizontal dimensions.
Results: Refraction was correlated with axial length (p<0.001).
Average foveal dimensions in the vertical and horizontal meridians,
respectively, were: pit depth, 134.6 and 128.5 micron; maximum
thickness, 345.7 and 338.6 micron; FWFM, 2135.0 and 2372.6
micron; FWHM, 820.0 and 883.3 micron (all p<0.001 in paired
t-tests). The fovea was thus wider horizontally by a factor of 1.11
at the maximum thickness and by 1.08 at half the pit depth. At
the maximum thickness, the asymmetry in pit dimensions did not
correlate with axial length, but the vertical/horizontal ratio of FWHM
pit width significantly decreased with axial length (p=0.04).
Conclusions: The foveal pit is wider horizontally than vertically and
the foveal pit depth is greater vertically. These asymmetries may be
related to the greater retinal thickness in the vertical meridian due to
the presence of the nerve fiber bundles. Alterations in the foveal pit
morphology in axially myopic eyes may be associated with retinal
stretching.
Commercial Relationships: Nancy J. Coletta, None; Rhett
Stroupe, None
Support: NIH grants R24 EY014817 and T35 EY007149
Automatic segmentation of the posterior vitreous boundary in
retinal optical coherence tomography
Alessio Montuoro, Sebastian M. Waldstein, Ana-Maria Glodan,
Dominika Podkowinski, Bianca S. Gerendas, Georg Langs, Christian
Simader, Ursula Schmidt-Erfurth. Christian Doppler Laboratory for
Ophthalmic Image Analysis, Vienna Reading Center, Department of
Ophthalmology, Medical University of Vienna, Vienna, Austria.
Purpose: Disorders of the vitreomacular interface (VMI) such as
vitreomacular traction and macular hole formation have recently
been made accessible to pharmacologic treatment by the introduction
of enzymatic vitreolysis. However, this therapeutic option is only
efficacious in a subset of patients with strictly defined patterns of
vitreous adhesions. Moreover, posterior vitreous detachment has been
demonstrated to impact the efficacy of intravitreally administered
antiangiogenic agents. Therefore, precise and reproducible
quantification and classification of the posterior vitreous boundary
and its adhesions at the macula is of major importance.
The aim of this study was to develop a method to fully automatically
segment the vitreous boundary in Spectral Domain - Optical
Coherence Tomography (SD-OCT) scans.
Methods: A set of 61 macula-centered SD-OCT volume scans from
patients available at the Vienna Reading Center was included. The
posterior vitreous boundary was manually annotated in 333 B-scans
(159/174 for training/testing).
A segmentation method was developed based on the concept of a
random forest trained on multiscale, rotation invariant eigenfeatures.
The local image orientation is estimated using the second order
central image moments and used to extract rotated windows around
each voxel. A principal component analysis was performed on these
windows and the resulting eigenvectors were used as features for a
random forest classifier.
The position of the vitreomacular interface was then detected by a
modified A* algorithm using the predicted probability map given by
the random forest as a cost function.
Results: Fully automated segmentation of the posterior vitreous
boundary was feasible in all included cases. The mean signed
distance between the expert annotation and the calculated
vitreomacular interface was -6.38 pixels (95% interval [-106, 77]).
Excluding areas where no expert annotation was present we find a
mean signed error of -11.327798 [-109.0, 7.0].
Conclusions: A fully automated segmentation method for the
detection of the posterior vitreous boundary in 3D SD-OCT was
developed and validated. This method may be used to automatically
determine patient eligibility for enzymatic vitreolysis and may
facilitate further studies of the vitreomacular interface in large-scale
antiangiogenic treatment trials.
Fig. 1: top: expert annotation, middle: automated segmentation,
bottom: raw B-scan
Commercial Relationships: Alessio Montuoro, None; Sebastian
M. Waldstein, None; Ana-Maria Glodan, None; Dominika
Podkowinski, None; Bianca S. Gerendas, None; Georg Langs,
None; Christian Simader, None; Ursula Schmidt-Erfurth, Alcon
(C), Bayer (C), Boehringer Ingelheim (C), Novartis (C)
Support: Austrian Federal Ministry of Science, Research and
Economy; National Foundation for Research, Technology and
Development.
Application of Normative Values to Improve Macular Asymmetry
Analysis in Glaucoma
Muhammed Alluwimi, Brett King, William H. Swanson. School of
Optometry, Indiana University, Bloomington, IN.
Purpose: Asymmetry analysis of macular thickness has been
introduced to clinical practice for evaluating damage to the macular
region, but without normative values. We previously derived
normative values for macular asymmetry, and found high betweensubject variability in control subjects which limits capability to detect
early macular damage. We have demonstrated that this variation can
be reduced by focusing on specific regions closer to the fovea. Here
we evaluate the potential of asymmetry analysis in detecting macular
damage in patients with glaucoma.
Methods: Twenty patients with glaucomatous visual field (242) defects (ages 40 to 85 years) and 30 age-similar controls were
recruited from a longitudinal glaucoma study. Subjects were imaged
with Spectralis OCT (V 5.4, Heidelberg Engineering, GmbH)
using the posterior pole protocol. A 64-cell grid is superimposed
on a 24°×24° area of the macula, and was manually adjusted to be
centered on the fovea; the central line of this grid was oriented in
the same direction as the foveal-disc angle. We demarcated 3 zones
per hemifield within the central ±8°, where we had found the lowest
variability. Asymmetry was calculated as the difference between the
average thicknesses of a superior zone and the corresponding inferior
zone. In order to validate our findings, high-density scans were used
with 10 of the 20 patients to derive en face images of the RNFL
bundles at different distances from the inner limiting membrane.
Results: Out of 180 zones in the control subjects, 2 zones were
flagged at p < 0.05, and one zone at p < 0.01, but none of 30 controls
were flagged with 2 contiguous zones. For patients, asymmetry
analysis showed 15 of 20 patients with at least 2 contiguous zones at
p < 0.05. RNFL damage seen in en face images confirmed damage in
the macular zones flagged by the asymmetry analysis.
Conclusions: By focusing on the central ±8°, we improved the
capability of the asymmetry analysis to detect macular thickness
reduction in patients with glaucoma. Our results were confirmed with
en face images of the RNFL bundles.
Corresponding glaucomatous damage was seen in the macular
asymmetry plot, en face image and visual field (24-2 and 10-2) gray
scale.
Commercial Relationships: Muhammed Alluwimi, None;
Brett King, None; William H. Swanson, Carl Zeiss Meditec (C),
Heidelberg Engineering (C)
Support: 1R01EY024542-01, EY007716
Reliability and Interchangeability of two Spectral-Domain
Optical Coherence Tomography in Healthy and Diseased Eyes
Mariam Fawzy1, 2, Jessica Taibl1, 3, Hang M. Tran1, 4, Samir I.
Sayegh1. 1The EYE Center, Champaign, IL; 2Spartan Health Sciences
University, Vieux Fort, Saint Lucia; 3University of Illinois UrbanaChampaign, Urbana, IL; 4Western University of Health Sciences,
Pomona, CA.
Purpose: Different SD-OCT systems yield different results in
measuring retinal nerve fiber layer thickness. We tested the influence
of macular edema on the difference in retinal thickness using two
different SD-OCT (RTVue Optovue 6.2 and Spectralis Heidelberg
5.7.5) and assessed their reliability.
Methods: 96 scans from 54 participants were included in this
observational clinical study with a breakdown of 34 normal eyes,
35 diabetic eyes with diabetic macular/retinal changes (DME, PDR,
and BDR), 20 eyes with dry/wet AMD, and 7 eyes with macular
edema not related to diabetes. 30 were females (56%) and 24 were
males (44%) aged 59 +/-21 years old. Scans were excluded if a major
artifact was present or if the scan or ETDRS measurement scale was
not centered on the fovea and if high myopia was present (D≥-10.00)
as the retinal nerve fiber layer is significantly thinner in severe
myopia.
Patients were imaged with both SD-OCT on the same clinic visit by
the same operator and readings were displayed using the ETDRS
scale. Patients were categorized into two group according to the
cutoff value of macular edema using Heidelberg 318um .15 scans
were above the cutoff. In 41 out of 54 participants, both eyes were
scanned to test for intra-reliability in both SD-OCT using the Intraclass coefficient (ICC)
Results: Using the RTVue ICC values were 0.97 and 0.82 for healthy
and pathological eyes, respectively. With Spectralis the values
were 0.98, and 0.87. A single factor ANOVA revealed a statistically
significant influence on the mean difference (MD) using the two
SD-OCT when Significant Macular Edema was present. R squared
revealed a 16% interaction level between macular edema and mean
difference.
Conclusions: Reliability of both SD-OCTsystems was influenced
by the presence of anatomical disruption in the retina. Significant
macular edema should be taken into special consideration when
applying an equation to use the two devices interchangeably.
Commercial Relationships: Mariam Fawzy, None; Jessica Taibl,
None; Hang M. Tran, None; Samir I. Sayegh, None
Support: ARVO-Egypt Travel Grant 2015
Automatic quantification of subretinal fluid in central serous
chorioretinopathy in 3D optical coherence tomography images
Freerk G. Venhuizen1, 2, Myrte Breukink2, Bram V. Ginneken1, Mark J.
Van Grinsven1, 2, Bart Bloemen1, 2, Carel C. Hoyng2, Thomas Theelen1,
2
, Camiel J. Boon3, Clarisa Sánchez1. 1Diagnostic Image Analysis
Group, Radboud University Medical Center, Nijmegen, Netherlands;
2
Department of Ophthalmology, Radboud University Medical Center,
Nijmegen, Netherlands; 3Department of Ophthalmology, Leiden
University Medical Center, Leiden, Netherlands.
Purpose: Central serous chorioretinopathy (CSC) is an ocular
disorder characterized by serous retinal detachment and associated
with fluid accumulation beneath the retina. Obtaining accurate
measures on the size and volume of the fluid deposit may be an
important biomarker to assess disease progression and treatment
outcome. We developed a system for automatic volumetric
quantification of subretinal fluid in optical coherence tomography
(OCT).
Methods: OCT images obtained from 15 patients with varying
presence of subretinal fluid were selected from the clinic. A 3D
region growing based algorithm was developed to segment the fluid
after selecting an arbitrary seed point located in the fluid deposit. The
obtained total volume, and the extent of the segmented fluid volume
were compared to manual delineations made by two experienced
human graders.
Results: A high intra-class correlation coefficient (ICC) value (0.997)
was obtained when comparing the fluid volume calculated by the
proposed method with the volume delineated by the two graders.
Similarly, the spatial overlap agreement on the obtained volumes,
measured with the Dice similarity coefficient (DC), between the
manual delineations and the software output was high (DC=0.87) and
comparable to the overlap agreement between observers’ delineations
(DC=0.85). In addition, the quantification time was reduced
substantially by a factor of 5 compared to manual assessment. The
quantified values obtained by the algorithm were shown to be highly
reproducible, obtaining a DC value of 0.99 and an ICC value of 0.98
when varying the seed point used for initializing the algorithm.
Conclusions: An image analysis algorithm for the automatic
quantification of subretinal fluid in OCT images of CSC patients was
developed. The proposed algorithm is able to accurately quantify the
extent of fluid deposits in a fast and reproducible manner, allowing
accurate assessment of disease progression and treatment outcome.
left: Fluid segmentation shown on B-scan, right: Fluid projected on
2D fundus image
3D representation of segmented fluid deposit
Commercial Relationships: Freerk G. Venhuizen, None; Myrte
Breukink, None; Bram V. Ginneken, None; Mark J. Van
Grinsven, None; Bart Bloemen, None; Carel C. Hoyng, None;
Thomas Theelen, None; Camiel J. Boon, None; Clarisa Sánchez,
None
A Multi-vendor Dataset and Standardized Evaluation
Framework for Retinal Cyst Segmentation
Christian Simader1, 2, Jing Wu2, Ana-Maria Glodan2, 1, Sebastian M.
Waldstein2, 1, Bianca S. Gerendas2, 1, Georg Langs3, 4, Ursula SchmidtErfurth2, 4. 1Vienna Reading Center, Department of Ophthalmology,
Medical University of Vienna, Vienna, Austria; 2OPTIMA Lab,
Department of Ophthalmology, Medical University of Vienna,
Vienna, Austria; 3Computational Image Analysis and Radiology Lab,
Medical University of Vienna, Vienna, Austria; 4Medical Imaging
Cluster, Medical University of Vienna, Vienna, Austria.
Purpose: Optical coherence tomography (OCT) has become
the most frequently used imaging method for retinal diseases. In
clinical practice physicians evaluate pathologic OCT morphologies,
e.g. cysts, as a basis for diagnostic decisions. However, precise
monitoring of disease progression and treatment success is difficult
due to a lack of validated automated segmentation algorithms for
these 3D OCT morphologies. We present a multi-vendor ground
truth dataset for training and testing of automated cyst segmentation
algorithms to be made publically available as part of a segmentation
grand challenge.
Methods: The presented dataset contains 16 OCT scans from 2
major scanners (Cirrus, Spectralis), divided into 3 sets. The first set
contains 8 scans (4+4) for algorithm training, annotated manually
by 2 expert graders. The other 2 sets were annotated by 1 grader: 4
scans for algorithm testing and another 4 for future challenge testing.
Annotations were performed in the B-scan plane using a proprietary
system and stored as separate 3D ground truth data.
Two accuracy measures are defined for objective comparison of
algorithm results to ground truth. One examines the overlap between
system results and ground truth based on the Sørensen–Dice index.
The second, based on Hausdorf distance, examines the distance
between the system cyst boundaries and ground truth. In addition
algorithm performance is computed based on 2 criteria, clinical
significance and size of cyst.
Results: Manual annotations for ground truth of the training set by 2
graders presented:
A mean number of annotated cysts per scan of 147 and 549, with a
mean grader difference of 8±11 and 22±32 for Spectralis & Cirrus
scans respectively. Further analysis showed that disparity between
cyst annotations only occurred in cases of high number of small cysts
per scan.
Conclusions: First manual annotations for ground truth training data
for cyst detection algorithms gave us similar results for both graders.
The difference in cyst numbers between vendors directly reflects the
lower number of B-scans in the Spectralis protocol. Annotations for
the next 2 vendors are being completed and will double our database.
The ability to objectively rank the currently expected 15 different
detection algorithms shall allow a better understanding of positive
and negative aspects of each method and improve performance of
retinal disease progression and treatment success monitoring.
Commercial Relationships: Christian Simader, Novartis (F); Jing
Wu, Novartis (F); Ana-Maria Glodan, Novartis (F); Sebastian M.
Waldstein, None; Bianca S. Gerendas, None; Georg Langs, None;
Ursula Schmidt-Erfurth, None
Support: Austrian Federal Ministry of Science, Research and
Economy, Christian Doppler Laboratory for Ophthalmic Image
Analysis
Reliability of a Manual Procedure for Marking the EZ Band
Endpoint Location
Rithambara Ramachandran1, Cindy Cai1, Dongwon Lee1, Benjamin
Epstein1, Kirsten G. Locke2, David G. Birch2, Donald Hood1,
3 1
. Psychology, Columbia University, New York, NY; 2Retina
Foundation of the Southwest, Dallas, TX; 3Ophthalmology, Columbia
University, New York, NY.
Purpose: The ends of the ellipsoid zone (EZ) band, aka IS/
OS border, on frequency domain optical coherence tomography
(fdOCT) holds promise as an effective measure for tracking disease
progression in clinical studies of retinitis pigmentosa (RP).[1-4] Here
we assess the intra- and inter-grader agreement for marking the EZ
band endpoint location (EPL).
Methods: After a training session, 3 graders (1 experienced, 2
inexperienced) marked the EPLs on a random sample of horizontal
line scans (Spectralis) from 45 RP patients. The average mean
deviation of their 30-2 visual field was -22.5 dB (range -2.2 to
-33.6dB), and their log MAR ranged from 0 to 0.8. The graders
marked the EPLs on the same scans 1 month later. To assess intraand inter-grader variability, intra-class correlation coefficients (ICC)
and signed and absolute differences between the EPL marked by
one grader and the average EPL marked by the other two graders
[MEAN(ΔEPL)] were calculated. These results were compared to the
EPLs marked by an automated segmentation algorithm.[5]
Results: Intra-grader agreement was very good (ICC, 0.99) and
the marked EPLs were not statistically significant different (paired
t-test) between the two times for any of the graders. The mean signed
difference between the two times was 0.07° ± 0.54° and the mean
absolute difference was 0.28° ± 0.46°. Likewise, the EPLs did not
differ significantly across graders and the mean ICC values were
0.98 (time 1) and 0.97 (time 2). The average signed MEAN(ΔEPL)
was 0.01° ± 0.57° and averaged absolute MEAN(ΔEPL) was 0.35°
± 0.45°. The automated algorithm correctly identified the EZ band
in only 33 of 45 scans. In 12 scans, the algorithm segmented the
proximal retinal pigment epithelium border instead. For the 33
correctly segmented scans, the mean signed and absolute difference
between the automated and marked EPL was -0.03° ± 1.40° and 0.93°
± 1.04° respectively.
Conclusions: With training, the inter- and intra-grader reproducibility
of manually marked EPLs of the EZ band was excellent. The EPL
confidence intervals, about 2°, are smaller than the 6° spacing of the
typical visual field used. For clinical studies, the currently available
automated segmentation algorithms, if used, will need to be corrected
manually.
1. Hood et al., BOE, 2011; 2. Birch et al., JAMA, 2013; 3.
Ramachandran et al., TVST, 2013; 4. Cai et al., IOVS, 2014; 5. Yang
et al., BOE, 2011.
Commercial Relationships: Rithambara Ramachandran, None;
Cindy Cai, None; Dongwon Lee, None; Benjamin Epstein, None;
Kirsten G. Locke, None; David G. Birch, Acucela (C), Neurotech,
USA (C), QLT (C); Donald Hood, Topcon, In (F)
Support: NIH grant R01-EY09076 and grant FDR-02543 from
Foundation Fighting Blindness
Automatic count of outer retinal hyper-reflective bands in normal
and retinitis pigmentosa subjects
Koenraad A. Vermeer1, Jelena Novosel1, 4, Laurence Pierrache1,
3
, Caroline C. Klaver3, 5, L. I. van den Born2, Lucas J. van Vliet4.
1
Rotterdam Ophthalmic Institute, Rotterdam Eye Hospital,
Rotterdam, Netherlands; 2Rotterdam Eye Hospital, Rotterdam,
Netherlands; 3Department of Ophthalmology, Erasmus Medical
Centre, Rotterdam, Netherlands; 4Quantitative Imaging Group,
Dept. of Imaging Physics, Delft University of Technology, Delft,
Netherlands; 5Department of Epidemiology, Erasmus Medical Centre,
Rotterdam, Netherlands.
Purpose: Investigation of structural changes in the outer retina in
normal and retinitis pigmentosa (RP) subjects requires segmentation
of the individual retinal layers. For segmentation, the number of
discernible hyper-reflective bands needs to be determined. Here, we
present and evaluate a method to count the number of hyper-reflective
bands in the outer retina in normal and RP macular OCT volumes.
Methods: The number of discernible hyper-reflective bands in the
outer retina may vary due to disease or overlapping bands. The
bands were modelled by a Gaussian mixture model (GMM) with a
varying number of components. Model parameters were optimized
by maximum-likelihood estimation. The information complexity
(ICOMP) criterion was then used to select the best model and
consequently the number of bands was set to the selected model’s
number of components. ICOMP balances the quality-of-fit with
the interdependence of the parameters. A complex model is only
preference if it was needed to model the data, i.e. if the better fit is
not penalized due to model degeneracy.
To validate the detection, 2 normal and 2 RP patients were imaged
with a Spectralis OCT system (Heidelberg Engineering, Germany).
The macular scan protocol combined 37 B-scans composed of 512
A-lines of 496 pixels, with 35 times averaging. A medical doctor
manually annotated the number of bands visible in the outer retina on
every second B-scan.
Manual annotations were compared to automatic counts and the
precision and sensitivity were calculated per A-line.
Results: Examples of manual and automatic band counts are shown
in Figure 1. The results of a quantitative assessment are listed in
Table 1. The precision for one, two and three bands was 88, 81 and
88%, respectively. Further, the corresponding sensitivities were 95,
84 and 77%, respectively.
Conclusions: A method for counting the discernible outer retinal
bands was presented. The preliminary results indicate that the method
was able to correctly count the visible bands in over 80% of A-lines.
Further improvements of the results may be obtained by postprocessing.
Measurement of retinal thickness in optical coherence
tomography images in healthy and unhealthy retinas using a
computerized algorithm
David Nguyen-Tri1, Jocelyn Faubert1, William Seiple2, 3, Olga
Overbury1. 1École d’optométrie, Université de Montréal, Montreal,
QC, Canada; 2Lighthouse Guild, New York, NY; 3Schoold of
Medicine, New York University, New York, NY.
Purpose: Single optical coherence tomography (OCT) images are
generally noisy and it is difficult for an algorithm to rapidly and
accurately establish the boundary of anatomical landmarks in crosssectional images of the retina. Here, we present an image analysis
algorithm that allows the delimitation of anatomical structures within
the retina in OCT raster scan images and measurement of their
relative distance in healthy and unhealthy retinas.
Methods: OCT raster scan images were obtained with an Optos
OCT/SLO in healthy and in diseased eyes. A software algorithm
was developed in order to process the images and delineate the
vitreoretinal boundary and the retinal pigment epithelium. The
distance between these boundaries was then measured in order to
determine retinal thickness.
Results: The results show that the algorithm can reliably provide
an accurate estimate of anatomical boundaries in OCT/SLO images.
The results also demonstrate a good fit between the estimated
location of the landmarks and their actual location as established by a
human observer in both healthy and diseased retinas. In raster scans,
individual thickness scans can be used to interpolate the position
of retinal landmarks between scan lines and obtain a more detailed
estimate of the position of retinal landmarks. These extrapolated
boundary positions can be used to estimate retinal thickness and
volume.
Conclusions: The algorithm provides a robust and valid method for
delimiting the boundaries of retinal anatomical structures in both
healthy and pathological eyes. This can be used to establish retinal
thickness an measure volume in healthy and diseased eyes.
Figure 1: Examples of manual and automatic counts of the visible
hyper-reflective bands in a macula-centered en-face image for a
normal and an RP subject.
OCT/SLO raster scan image with location of retinal landmarks as
estimated by the computerized algorithm. The red line represents
the estimated location of the vitreoretinal interface. The blue line
represents the estimated location of the RPE.
Commercial Relationships: David Nguyen-Tri, None; Jocelyn
Faubert, None; William Seiple, None; Olga Overbury, None
Table 1: Confusion matrix of manual and automatic counts of visible
bands (one, two and three). The numbers in the table correspond to
the number of A-lines in all eyes.
Commercial Relationships: Koenraad A. Vermeer, None; Jelena
Novosel, None; Laurence Pierrache, None; Caroline C. Klaver,
None; L. I. van den Born, None; Lucas J. van Vliet, None
Support: Stichting Ooglijders
Relationship between Axial length and Macula Retinal layer
thickness in Normal Human Subjects
Kwame Antwi-Boasiako1, Jonathan D. Oakley2, Daniel Russakoff2,
Sherine John1, Nimesh B. Patel1. 1College of Optometry, University
of Houston, Houston, TX; 2Voxeleron LLC, California, CA.
Purpose: Optical coherence tomography (OCT) determined
thickness measures of the macula are often used for the diagnosis and
management of retinal and optic nerve pathology. Current methods
use a fixed angular projection volume scan to derive thickness
measures. The purpose of this study was to assess the relationship
between axial length and the spectral domain OCT-measured
thickness of seven retinal layers in normal human subjects after
incorporating transverse scaling.
Methods: For each of 98 normal subjects recruited, one randomized
eye was enrolled in the study. After pupil dilation with 1%
tropicamide, a 49 line, 20 × 20 degree OCT raster scans, centered on
the fovea, were acquired on each subject with averaging of 16 frames.
Corneal curvatures and axial length parameters were measured
with an optical biometer. Using an automated retinal segmentation
algorithm (Orion, Voxeleron LLC, Pleasanton, CA), average
thickness, ETDRS measures were exported for seven retinal layers.
The average thicknesses of these layers at the three annulus diameters
of the macular grid (2mm, 3mm and 5mm annulus rings) were
correlated with axial length measurements using linear regression
analysis.
Results: The average age of subjects recruited was 26.14 years (range
21 to 42 years). All subjects had healthy retinas and optic nerve
heads as observed with fundus photography. The mean axial length
was 24.60 mm (range 22.02 to 27.72mm). There was a significant
relationship between scan focus and axial length (slope = -0.35D/mm,
r2=0.5067, p<0.01). Based on instrument scaling that incorporates
scan focus and corneal curvature, there was a statistically significant
relationship between axial length and the thickness measures of the
inner nuclear layer (p<0.01) of the outer annulus. However, only
a small percentage of the variance (r2 < 0.108) could be accounted
for by the linear regression. No significant correlations were found
between axial length and average thickness measures from annuli for
the other six retinal layers (Bonferroni correction, p > 0.017).
Conclusions: When scans are scaled, there was a weak relationship
between axial length and inner nuclear layer thickness in the
parafoveal region (3-5mm annulus diameter). This relationship might
not be clinically significant based on the variance. Hence, when
scaling is incorporated into thickness analysis of the macula region,
the effects of ocular magnification are minimal.
Commercial Relationships: Kwame Antwi-Boasiako, None;
Jonathan D. Oakley, Voxeleron LLC (E), Voxeleron LLC (P);
Daniel Russakoff, Voxeleron LLC (E), Voxeleron LLC (P); Sherine
John, None; Nimesh B. Patel, None
Support: K23 EY021761, P30 EY007551
Validity of an automated algorithm for Choroidal Volume
measurement
Jay Chhablani1, Ashutosh Richhariya1, Soumya Jana2, Kiran Kumar2,
Srinath Nizampatnam2. 1Vitreo-retina, L V Prasad Eye Institute,
Hyderabad, India, Hyderabad, India; 2Electrical Engg, Indian Institute
of Technology, Hyderabad, India.
Purpose:
Manual segmentation of choroidal boundaries is time consuming,
tedious and may have inter observer variation. We developed an
automated algorithm to segment choroid and provide choroidal
volume measurements. Present study aims to validate the automated
algorithm in comparison to expert manual segmentation.
Methods: Three sets of 97 spectral domain optical coherence
tomography (SD-OCT) scans, obtained using enhanced depth
imaging, were analyzed. Automated segmentation algorithm to
detect the upper and lower boundaries of the choroid, was based on
structural similarity, adaptive Hessian analysis and tensor voting.
Single observer performed manual segmentation twice, in a masked
fashion, to obtain intra-observer repeatability. Automated algorithm
was validated using 291 SD-OCT scans from three 97-scan datasets
as well as individual sets. Cross correlation coefficient and Dice
coefficient was used to correlate manual and automated segmentation.
Results: Overall correlation coefficient and Dice coefficient
between manual segmentation was 99.77± 0.19% and 96.73 ± 1.39%
respectively. Overall correlation coefficient and Dice coefficient
between average manual segmentation was 99.49 ± 0.41% and
93.25 ± 3.07% respectively. Correlation coefficients and absolute
differences in choroidal volume between manual and automated
segmentation for three sets are shown as Table 1 and 2 respectively.
Conclusions: Automated algorithm was in close agreement with
manual segmentation with average correlation coefficient of about
99% and mean Dice coefficient of about 93%.
Comparison of cross correlation coefficient and Dice coefficient
between automated and manual segmentation, considering three
individual datasets of 97 scans each and all 291 scans from three
datasets
Comparison of absolute differences in choroidal volume obtained
from automated and manual segmentations
Commercial Relationships: Jay Chhablani, None; Ashutosh
Richhariya, None; Soumya Jana, None; Kiran Kumar, None;
Srinath Nizampatnam, None
Error rate of automated choroidal segmentation using sweptsource optical coherence tomography
Mingui Kong, Doo-ri Eo, Gyule Han, Sung Yong Park, Don-Il Ham.
Ophthalmology, Samsung Medical Center, Sungkyunkwan University
School of Medicine, Seoul, Korea (the Republic of).
Purpose: To investigate the error rate of automated choroidal
segmentation and the effect of frame-averaging on error rate
Methods: A horizontal b-scan at the fovea was performed in
patients having various retinochoroidaldisorders using sweptsource optical coherence tomography (OCT) with frame-averaging
technique. Scanned images were classified into four morphological
groups;normal from fellow eyes (NF), normal from pathologic
eyes (NP), retinal abnormalty (R), and retinochoroidal abnormality
(RC) group, respectively.Choroidal segmentation was automatically
performed using built-in software of a swept-source OCT device, and
the error rate of choroidal segmentation was analysed.
Results: Qualified images for all four averaging types with different
number of averaged frames were acquired in 89 eyes of 77 patients.
Images of 12, 20, 24, and 33 eyes were classified as NF, NP, R,
and RC group, respectively. The choroidal segmentation error was
detected in 1-2 images (8.3-16.7%) in the NF group, 3-6 images
(15.0-30.0%) in the NP group, 4-8 images (16.7-33.3%) in the R
group, and 17-19 images (51.5-57.6%) in the RC group.The error
rate was significantly higher in RC group than other groups (P<0.05).
Increasing the number of frames for averaging showed no significant
effect on the error rate in all groups (P>0.05).
Conclusions: Automated choroidal segmentation showed a high
error rate in images with choroidal abnormalities, and the averaging
effect could not reduce the error rate significantly. Thus, further
technological improvement is needed to increase the accuracy of the
automated choroidal segmentation.
Commercial Relationships: Mingui Kong, None; Doo-ri Eo, None;
Gyule Han, None; Sung Yong Park, None; Don-Il Ham, None
Choroidal Cross Section Area and Choroidal Volume and Axial
Length
Jost B. Jonas1, Leonard Holbach2, Songhomitra Panda-Jonas1.
1
Heidelberg University, Germany, Mannheim, Germany; 2Department
of Ophthalmology, Erlangen, Germany.
Purpose: To examine whether the choroidal cross sectional area and
estimated choroidal volume are associated with a longer axial length
in human eyes.
Methods: Histologic anterior-posterior sections running through
the pupil and the optic nerve head were examined. Using a light
microscope, we measured thickness of the choroid at the limbus,
ora serrata, equator, midpoint between equator and posterior pole
(MPEPP), peripapillary region and posterior pole. The axial length
and the horizontal and vertical diameters of the globe were measured
with a ruler or caliper. We calculated the choroidal volume as inner
scleral surface area times choroidal thickness.
Results: The histomorphometric study included 53 human globes (53
subjects; mean age: 60.7 ± 19.4 years) enucleated due to malignant
choroidal melanoma or due to secondary angle-closure glaucoma.
Mean axial length was 27.1 ± 4.0 mm (median: 26.0 mm; range:
21-39 mm). Choroidal thickness measurements decreased with
increasing axial length for values taken at the midpoint between
posterior pole and equator (P=0.08;r=-0.24), at the optic nerve head
border (P=0.02;r=-0.33) and at the posterior pole (P=0.008;r=0.36;beta:-0.36). Choroidal thickness measured at the equator
(P=0.58) and at the ora serrata (P=0.66) were not associated with
axial length. Neither choroidal cross section area measured from
the ora serrata to the posterior pole (P=0.58) nor choroidal volume
measured from the ora serrata to the posterior pole (P=0.35) were
significantly associated with axial length.
Conclusions: Choroidal thickness in the posterior quarter of the
globe decreased with increasing axial length, most marked at the
posterior pole. Total choroidal volume was not associated with
axial length, pointing against a choroidal volume enlargement to
play a role in, or to be associated with, the process of myopic axial
elongation.
Commercial Relationships: Jost B. Jonas, Alimera Co (C),
Allergan Co (C), Boehringer Ingelheim Co (C), Merck Sharp &
Dohme Co., Inc (C), Patent holder with CellMed AG, Alzenau,
Germany (P), Pfizer Co (C), Sanofi Co (C); Leonard Holbach,
None; Songhomitra Panda-Jonas, None
Temporal choroidal hump in the Swept-Source OCT choroidal
thickness profile of a healthy population
Jorge Ruiz-Medrano1, Ignacio Flores-Moreno2, Javier A Montero3,
Jose M Ruiz-Moreno2. 1Hospital Universitario Clinico San Carlos,
Madrid, Spain; 2Department of Ophthalmology, Castilla La Mancha
University. Albacete. SPAIN, Albacete, Spain; 3Pío del Río Hortega
University Hospital, Ophthalmology Unit. Valladolid. SPAIN,
Valladolid, Spain.
Purpose: The choroid and its involvement in a variety of ocular
pathologies has been the subject of intensive study during the last few
years. The purpose of this study is to describe choroidal thickening
temporal to the fovea in the horizontal macular choroidal thickness
(CT) profile in a healthy population using swept-source optical
coherence tomography (SS-OCT).
Methods: Cross-sectional, non-interventional study. 276 eyes from
154 healthy patients, with spherical equivalent (SE) ±3D were
scanned using a SS-OCT (Topcon Corporation, Japan). The presence
of choroidal thickening (hump) in the analysis of the horizontal
macular CT profile was recorded and the distance to the fovea and
CT was manually measured. The hump was defined as an evident
choroidal thickening compared to the two adjacent measured points
by SS-OCT without any other relevant choroidal finding.
Results: Choroidal hump was found in 36 eyes (12.8%). All the
humps were found in the temporal sector, with a mean distance to the
fovea of 4427.3±627.9 mm (range 2531 to 5492) with a mean hump
CT of 372.1±76.8 mm (range 168 to 538). Mean age of the patients
with hump was 16±19 years (range 4 to 82) vs. 36±25 years (range
3 to 95) in the patients without hump (p=0.000; Student t test for
unpaired data). A thicker temporal CT was noted in the hump group.
Conclusions: Temporal choroidal humps can be considered normal
variations without clinical significance. This focal increase in
temporal CT is more frequent in younger patients, and it can be
explained by the reduced scleral stiffness in younger age and the
traction applied by the insertion of the inferior oblique muscle.
Commercial Relationships: Jorge Ruiz-Medrano, None; Ignacio
Flores-Moreno, None; Javier A Montero, None; Jose M RuizMoreno, None
4D longitudinal choroidal thickness quantification improves
reproducibility
Ipek Oguz, Li Zhang, Kyungmoo Lee, Michael D. Abramoff, Milan
Sonka. The University of Iowa, Iowa City, IA.
Purpose: Accurate and reproducible measurement of choroidal
thickness is important for longitudinal studies of disease mechanisms
and management. Joint segmentation of multiple 3D scans over time
(termed 3D+time or 4D) has the potential to improve accuracy.
Methods: Topcon SS OCT scans were acquired every 2 weeks during
anti-VEGF induction (7 scans/subject). Bruch’s membrane (BM) was
segmented using the Iowa Reference Algorithms (www.iibi.uiowa.
edu/octexplorer) to flatten the images. In the 3D approach, upper
and lower surfaces of the choroid were segmented for each scan
independently, using a graph-based algorithm. In the 4D approach,
the 7 scans were segmented jointly by linking the individual graphs
based on temporal registration. These links encourage temporal
coherence of the segmentation while allowing up to 52μm surface
position variation between each successive timepoint pair. Paired
t-tests were used to compare choroidal thickness measurements
obtained from the 3D and 4D approaches.
Results: 15 patients with exudative AMD were included.
Compared to independent 3D segmentations, choroidal thickness
was substantially more reproducible in 4D (standard deviation of
thickness was 2.43μm in 4D vs. 7.05μm in 3D, p<0.001). The mean
thickness (50.6 vs. 49.2μm) and temporal thinning rate (4.28 vs.
7.43μm/12 weeks) were not significantly different between the two
methods (p=0.34, p=0.23), suggesting that the 4D approach is less
affected by measurement noise and not exhibiting measurement bias,
leading to increased temporal sensitivity.
Conclusions: We have developed the first automated 4D method for
jointly quantifying choroidal thickness in longitudinal OCT studies.
Our method is robust to image noise and produces more reliable
choroidal thickness measurements than a sequence of independent 3D
segmentations.
Fig. 1: Choroidal thickness computed in 3D and 4D. Colors
denote individual patients. 4D results show significantly smaller
measurement variability over time.
Fig. 2: Choroidal segmentation consistency and accuracy. (a) 4D
(red) and 3D (blue) segmentations generally agree with each other,
except for areas of high ambiguity (white rectangle). Note that the
3D approach erroneously cuts through choroidal vessels. (b,c) The
7 individual 3D segmentations show up to 280μm local differences,
while the 4D joint segmentation resolves these discrepancies and
offers segmentation consistency.
Commercial Relationships: Ipek Oguz, None; Li Zhang, None;
Kyungmoo Lee, None; Michael D. Abramoff, IDx LLC (C), IDx
LLC (I), University of Iowa (P); Milan Sonka, University of Iowa
(P)
Support: National Institutes of Health grants R01 EY019112, R01
EY018853 and R01 EB004640.
Factors influencing choroidal thickness measurements using
swept-source optical coherence tomography
Albert Caramoy, Bernd Kirchhof, Claus Cursiefen, Ludwig M.
Heindl. Department of Ophthalmology, University of Cologne,
Cologne, Germany.
Purpose: To study the variability of choroidal thickness in subjects
without macular pathology and its influencing factors using sweptsource optical coherence tomography.
Methods: In 123 volunteers without macular pathology sweptsource optical coherence tomography was performed (DRI OCT1, TopCon Corporation, Japan). The Early Treatment Diabetic
Retinopathy Study (ETDRS) grid was centered manually at the
fovea. The following retinal layers were delineated automatically:
RETINA=total retinal thickness from internal limiting membrane to
Bruch’s membrane, CSI=choroid sclera interface. The retinal layers
were analyzed using multiple regressions with respect to gender, age,
axial length and day time.
Results: Mean age was 33.40 ± 16.36 years. Forty five men and
78 women were analyzed. Mean RETINA was 285.85 ± 14.53 mm
and 287.18 ± 12.93 mm for the right and left eye respectively. Mean
CSI was 273.94 ± 77.77 and 271.19 ± 78.85 mm for the right and
left eye respectively. The assumptions of linearity, independence of
errors, homoscedasticity, unusual points and normality of residuals
were met. The RETINA correlated significantly for gender, age and
axial length (all p values < 0.05), however no significant correlation
was found with regard to day time (p=0.579). CSI correlated also
significantly for gender, age and axial length (all p values < 0.05), but
not significantly for day time (p=0.249). Among these values CSI was
only moderately negatively correlated with axial length (r= -0.44,
p<0.001).
Conclusions: Swept-source optical coherence tomography and
the use of 1,050 nm wavelength allow rapid measurements of the
choroid. Total retinal thickness and choroid sclera interface were
influenced by gender, age and axial length. No correlation with regard
to day time was found.
Commercial Relationships: Albert Caramoy, None; Bernd
Kirchhof, None; Claus Cursiefen, None; Ludwig M. Heindl, None
Interocular differences in choroidal thickness and volumen in
healthy eyes
Isabel Pinilla Lozano1, 2, Elvira Orduna Hospital3, Francisco J.
Segura5, 2, Carmen Lopez de Lafuente4, 2, Diana Perez-Garcia1,
2
, Nicolas Cuenca7, Emilio Abecia6, Ana I. Sanchez Cano4, 2.
1
Ophthalmology, Universitary Hospital Lozano Blesa, Zaragoza,
Spain; 2IIS Aragon, Zaragoza, Spain; 3Department of Physiology,
School of Medicine, Zaragoza University, ., Zaragoza, Spain;
4
Department of Applied Physics, Zaragoza University, Zaragoza,
Spain; 5Department of Surgery, School of Medicine, Zaragoza
University, Zaragoza, Spain; 6Ophthalmology, Hospital Unviersitario
Miguel Servet, Zaragoza, Spain; 7Department of Physiology Genetics
and Microbiology, Alicante University, San Vicente del Raspeig,
Alicante, Spain.
Purpose: To analyze and establish the interocular limits in choroidal
thickness (CT) and volume (CV) in a healthy young adult population.
Methods: One hundred and twenty eyes of 60 healthy subjects
underwent optical coherence tomography (OCT) with enhanced
depth imaging (EDI). CTs and CVs at the three rings (1, 3 and 6mm)
of the defined areas of the Early Treatment Diabetic Retinopathy
Study (ETDRS) were measured manually. Topographic variation and
relative interocular differences in CT and CV were analyzed.
Results: The mean±SD subfoveal CT in the right eye (RE) and
the left eyes (LE) were 338.33±73.82mm and 357.80±76.11mm
respectively, with a high correlation between the two eyes
(r=0.826). Subfoveal CV were respectively, 0.27±0.06mm3 and
0.28±0.06mm3 (r=0.821). In the 3mm and 6mm rings, mean CT (RE
vs LE) were 332.08±69.41mm vs 346.90±70.73mm (r=0.860) and
306.28±58.30mm vs 313.90±58.93mm (r=0.901). Regarding CV, they
were 2.37±0.52mm3 vs 2.46±0.50mm3 (r=0.855) and 8.84±1.72mm3
vs 9.12±1.74mm3 (r=0.894). Although relative interocular difference
is not significant, absolute interocular differences in subfoveal CT
and CV may reach 95mm and 0.07mm3 respectively, 74mm and
0.62mm3 in a 3mm ring and 59mm and 1.85mm3 in a 6mm ring.
Conclusions: In healthy eyes there was a no significant relative
interocular difference in CT or CV but normal absolute interocular
differences should be established to differentiate healthy eyes from its
pathological fellows.
Commercial Relationships: Isabel Pinilla Lozano, None; Elvira
Orduna Hospital, None; Francisco J. Segura, None; Carmen
Lopez de Lafuente, None; Diana Perez-Garcia, None; Nicolas
Cuenca, None; Emilio Abecia, None; Ana I. Sanchez Cano, None
Support: PI13/01124 grupo DGA B99
Choroidal And Retinal Layer-By-Layer Analysis In Intermediate
AMD Patients Using Manually-Segmented SD-OCT High
Resolution B-Scans
Andrea M. Coppe, Giuliana Lapucci, Guido Ripandelli.
Ophthalmology, GB Foundation Study Ophth IRCCS, Rome, Italy.
Purpose: To study the retinal and choroidal structure in age-related
macular degeneration (AMD) patients with bilateral large drusen and
pigment changes (intermediate AMD [iAMD]) analyzing SD-OCT
high resolution (HR) B-scans
Methods: 21 eyes of 42 patients with iAMD, mean age 77.90±8.60
years, 11F and 10M, were studied. 19 eyes randomly selected in 38
healthy, age-matched, subjects were used as control eyes (CEs).
All subjects had ametropia < 3D and underwent clinical examination
including visual acuity evaluation using ETDRS chart and slit-lamp
biomicroscopy with +90 D lens. The macula was studied with a
SD-OCT Cirrus 5000 (Carl Zeiss Meditec, Inc) using a HR 6 mm
horizontal B-scan (EDI HD5 line raster merged in a single scan),
crossing the center of the fovea. The B-scan was divided in the
middle of the fovea and the nasal and temporal scans were analyzed
separately; total retina (TR) from inner limiting membrane to Bruch’s
membrane, subretinal layers outer nuclear (ONL), inner nuclear
(INL), ganglion cells (GCL) and choroid (Ch) were segmented
using the software ImageJ ver.1.48v. Data underwent to indipendent
2-sample unequal variance t-test to evaluate difference between
groups; a level of P<0,05 was accepted as statistically significant
Results: No significant differences were found when nasal and
temporal ONL, INL and GCL thickness were compared in iAMD
eyes vs CEs (0.379±0,005 vs 0.382±0.007 mm2, P=0.70; 0.370±0,007
vs 0.366±0.006 mm2, P=0.63; 0.153±0.004 vs 0.158±0.004 mm2,
P=0.37, 0.135±0.003 vs 0.143±0.004 mm2, P=0.09; 0.226±0.005
vs 0.218±0.003 mm2, P=0.21, 0.218±0.004 vs 0.218±0.004
mm2, P=0.93; respectively). Significant differences were found
between Ch and TR thickness (0.855±0.062 vs 0.680±0.054 mm2,
P=0.04; 0.901±0.040 vs 0.727±0.053 mm2, P=0.01; 0.918±0.015
vs 0.984±0.029 mm2, P=0.05; 0.814±0.008 vs 0.915±0.041 mm2,
P=0.02; respectively)
Conclusions: Data from this study show that SD-OCT HR B-scans
analysis with proper software allows a reliable retinal and choroidal
segmentation. We found a significant choroidal and total retinal
thinning in iAMD eyes, as also demonstrated by other studies;
however, it is interesting to highlight that in this study ONL, INL
and GCL thickness differences between iAMD eyes and CEs were
not significant. These findings could support the hypothesis that the
choroid could play major role in the pathogenesis of iAMD
Commercial Relationships: Andrea M. Coppe, None; Giuliana
Lapucci, None; Guido Ripandelli, None
Validation of Segmentability Index for Automated Prediction of
Segmentation Reliability in SD-OCT Scans
Kyungmoo Lee1, 2, Gabrielle H. Buitendijk3, 4, Hrvoje Bogunovic1, 2,
Henriet Springelkamp3, 4, Albert Hofman4, Andreas Wahle1, 2, Milan
Sonka1, 2, Johannes R. Vingerling3, 4, Caroline C. Klaver3, 4, Michael
D. Abramoff5, 1. 1Electrical and Computer Engineering, University
of Iowa, Iowa City, IA; 2Iowa Institute for Biomedical Imaging,
University of Iowa, Iowa City, IA; 3Ophthalmology, Erasmus
Medical Center, Rotterdam, Netherlands; 4Epidemiology, Erasmus
Medical Center, Rotterdam, Netherlands; 5Ophthalmology and Visual
Sciences, University of Iowa, Iowa City, IA.
Purpose: Though some commercially available SD-OCT devices
provide an image quality index, these are not available on all, nor
are they consistent across all SD-OCT devices. More importantly,
they do not predict the reliability of different structures, such as
retinal layers, in the SD-OCT scan volume. We have developed a
new metric, Segmentability Index (SI), to predict the reliability of the
segmentation.
Methods: 3808 macular SD-OCT (3D OCT-1000, Topcon Europe,
Netherlands) volumes (512 × 128 × 650 voxels, 6.0 × 6.0 × 2.3 mm3)
were obtained from both eyes of 1128 subjects (74.7 ± 8.3 years,
41% male) randomly selected from the population-based Rotterdam
study. A subset consisting of 50 OCT volumes with successful RNFL
segmentations and 50 volumes with failed segmentations (Iowa
Reference Algorithms: https://www.iibi.uiowa.edu/content/iowareference-algorithms-human-and-murine-oct-retinal-layer-analysisand-display) was created. The SI is obtained from a random forest
regressor based on 12 features: aggregate on-surface costs of inner/
outer RNFL and outer RPE surfaces, mean and standard deviations
of whole OCT voxel intensities, and whole edge-based costs of darkto-bright and bright-to-dark transitions. A 10-fold cross-validation on
these 100 volumes was compared to those obtained from a random
forest regressor trained by a well-known quality index, maximum
tissue contrast index (mTCI) based on the OCT voxel intensity
distribution [1], using receiver operating characteristic (ROC)
analysis.
[1] Huang Y, Gangaputra S, Lee KE, et al. Signal quality assessment
of retinal optical coherence tomography images. Invest Ophthalmol
Vis Sci. 2012;53(4):2133-41.
Results: Areas under the curves (AUCs) of the mTCI and SI [95%
confidence intervals (CIs)] are 0.722 [0.623, 0.822] and 0.892 [0.828,
0.957], respectively (Fig. 1), so that the SI ROC curve is significantly
better than the mTCI curve (p < 0.001).
Conclusions: The SI is well suited to identify OCT scans that provide
successful automated RNFL segmentations and those that do not
allow reliable segmentations. Using the SI, scans with unreliable
segmentations can be identified allowing more reliable automated
segmentation analyses in population studies, in a device-independent
manner.
Figure 1. ROC graph of the mTCI and SI.
Commercial Relationships: Kyungmoo Lee, None; Gabrielle
H. Buitendijk, None; Hrvoje Bogunovic, None; Henriet
Springelkamp, None; Albert Hofman, None; Andreas Wahle,
None; Milan Sonka, University of Iowa (P); Johannes R.
Vingerling, None; Caroline C. Klaver, None; Michael D.
Abramoff, IDx LLC (C), IDx LLC (I), University of Iowa (P)
Support: NIH Grants R01-EY017066, R01-EY018853, R01EB004640; the Department of Veterans Affairs; the Netherlands
Organization for Scientific Research, the Hague; Swart van
Essen, Rotterdam; Bevordering van Volkskracht, Rotterdam;
Rotterdamse Blindenbelangen Association, Rotterdam; Algemene
Nederlandse Vereniging ter Voorkoming van Blindheid, Doorn, The
Netherlands; Oogfonds Nederland, Utrecht; MDFonds, Utrecht;
Vereniging Trustfonds Erasmus Universiteit Rotterdam, Rotterdam,
The Netherlands; and Lijf en Leven, Krimpen aan de IJssel, The
Netherlands; Topcon Europe BV, Capelle aan den IJssel, The
Netherlands.
Methods: Eyes were imaged with a custom 1060-nm swept-source
OCT system. The 3D images were corrected for axial motion and
smoothed. Retinal nerve fiber layer (RNFL) was segmented using a
3D graph-cut algorithm.
Longitudinal data were registered by voxel-wise retinal surface
matching between the baseline (template) and each follow-up
image (target). The surfaces were first registered approximately by
mathematical current representation and reproducing kernel Hilbert
space norms. Spherical demons registration followed to yield voxelwise, exact matching.
For cross-sectional data, a common atlas was generated as a mean
template using the f-shape framework such that each surface could be
approximated as a deformation of the atlas. The mean template RNFL
surface geometry and thickness were concurrently optimized.
To characterize the repeatability of the registration pipeline, 3
healthy subjects were imaged 9 times over 3 weeks, and the RNFLs
of the follow-up images were registered to that of the baseline.
RNFL thickness was compared at each voxel across the time points
to quantify the variability. For cross-sectional analysis, mean
templates were generated from 20 normal eyes of 10 subjects and
26 glaucomatous eyes of 16 subjects, and the RNFL thickness was
compared at each voxel between the two groups.
Results: Fig. 1 shows an example of longitudinal RNFL thickness
maps acquired over 3 weeks, with the follow-up images registered
to the baseline image. Table 1 demonstrates the high voxel-wise
repeatability of the RNFL thickness measurement by registration.
Fig. 2 shows the average RNFL thickness maps of the normal and
glaucomatous eyes, generated by the f-shape framework, which show
the spatial pattern of glaucomatous RNFL thinning. Fig. 3 shows
the regional statistical significance in differentiating the normal
and glaucomatous groups. Fig. 2 and 3 suggest the RNFL thinning
in glaucoma may occur preferentially along the thickest part of the
RNFL.
Conclusions: We demonstrated computational frameworks capable
of spatially-detailed study of longitudinal or cross-sectional retinal
morphology and voxel-wise shape variability analysis.
Quantitative longitudinal and cross-sectional shape analysis of
retina by registration
Sieun Lee1, Morgan Heisler1, Paul Mackenzie2, Marinko V. Sarunic1,
Mirza F. Beg1. 1School of Engineering Science, Simon Fraser
University, Burnaby, BC, Canada; 2Department of Ophthalmology
and Visual Science, University of British Columbia, Vancouver, BC,
Canada.
Purpose: Quantitative shape analysis of multiple retinas is often
limited by the lack of spatial correspondence among the retinas. We
present two frameworks that register longitudinal and cross-sectional
optical coherence tomography (OCT) data and enable detailed
morphological comparison.
Commercial Relationships: Sieun Lee, None; Morgan Heisler,
None; Paul Mackenzie, None; Marinko V. Sarunic, None; Mirza F.
Beg, None
Support: National Science and Engineering Research Council,
Canadian Institute of Health Research, Michael Smith Foundation
for Health Research, Alzheimer Society Research Program, Brain
Canada
A fully automatic ex-vivo Brn3a retinal segmentation algorithm
describing the pattern of regional RGC loss in a rat ocular
hypertension model
Lawrence Langley, Ben Davis, Li Guo, Lisa Turner, Shereen Nizari,
M Francesca Cordeiro. UCL Institute of Opthalmology, London,
United Kingdom.
Purpose: Previously, researchers looking to quantitatively assess
RGC health were often required to manually define regions of
interest (ROIs) for assessment of Brn3a or fluorogold -labelled
retinal wholemounts, due to the impractibility and workload of
performing whole retinal counts. In the present study, a segmentation
algorithm is presented for the fully automatic determination of spatial
distribution of retinal ganglion cells (RGCs) in retinal wholemounts,
which enables the assessment of changes in RGC density and nearest
neighbour distance (NND). This is applied to a model of rat ocular
hypertension, where we test the assumption that initial RGC loss
occurs in the retinal periphery.
Methods: A segmentation algorithm was designed to divide
wholemount retinal images into 15 concentric rings or circles of
0.6mm increasing diameter. Each retina was further segmented into
pre-defined quadrants. The area of each segment was automatically
determined for each region, and RGC counts made using a previouly
described algorithm. Mean RGC density and NND was next
determined for each retinal area. This novel algorithm was applied to
wholemounts from a rodent OHT model at 3 and 8 weeks (minimum
n=3/time point, and compared to controls).
Results: The algorithm was easy to use and quick - the whole process
taking less than 5 min per retina. Global mean NND increased
significantly by 8.2 % between the control and 3 week OHT model
(p= 0.0052), with a significant reduction in global RGC density
(p= 0.0009). The whole circle and regional mean NND and density
graphs indicate initial loss of RGCs occurring in the peripheral
regions. No significant difference was found between the densities
of the 3 week post OHT and the control eyes in the central 2 circles.
However, the outer 2 circles of the retina displayed a significant
decrease in density (p= 0.0043).
Conclusions: The lack of significant difference between control and
3 week mean NND in the central rings indicates that in the initial
stages of OHT, RGC loss occurs predominantly in the periphery of
the retina. Central loss of RGCs in our OHT model occurred therafter.
This result supports the findings of existing studies on OHT, and
suggests that our segmentation algorithm is an effective tool that can
be applied to many other models involving retinal wholemount image
analysis.
Imaging rhodopsin molecular contrast in vivo by optical
coherence tomography
Shuliang Jiao1, Tan Liu1, Byron L. Lam2, Rong Wen2. 1Biomedical
Engineering, Florida International University, Miami, FL; 2Bascom
Palmer Eye Institute, Miami, FL.
Purpose: To develop and validate a novel visible OCT technology
for imaging the molecular contrast of rhodopsin in vivo. The ultimate
goal of the study is to provide an imaging tool to assess the functions
of rod photoreceptors.
Methods: Rhodopsin, the light-sensing molecule in the outer
segments of rod photoreceptors, is responsible for converting light
into neuronal signals in a process known as phototransduction. After
absorbing a photon, rhodopsin undergoes conformational changes
with a dramatic shift in its absorption spectrum: the maximum
absorption peak shifts from 500 nm to 380 nm. This absorption
change, known as rhodopsin photo-bleaching, provides a contrast
to image rhodopsin. We report a novel 3-D retinal densitometry
technique based on visible-light OCT for in vivo molecular imaging
of rhodopsin. The system uses a light source centered at 532 nm
closing to the peak absorption wavelength. With a bandwidth of 9.3
nm the system achieved a depth resolution of 13 micro-meters in air.
The depth resolution allows the visualization and segmentation of the
location where the absorption change occurs and provides an accurate
assessment of rhodopsin content.
Results: We have applied the technology to image Sprague Dawley
rats. The technology has successfully revealed the depth-resolved
rhodopsin distribution in the retina, and thus provides direct
visualization of the photon propagation pathways. We have designed
and successfully performed a pattern bleach experiment, the results
of which is shown in the figure. The calculated image clearly showed
the bleached strip of rhodopsin in both the projection image of the 3D
OCT data an the depth resolved cross-sectional images. The results
proved conclusively that our technology is able to image rhodopsin
distribution in vivo.
Conclusions: Rhodopsin VIS-OCT can be used to quantitatively
image rhodopsin distribution and thus assess the distribution of
functional rod photoreceptors in the retina. Rhodopsin VIS-OCT can
be adapted into a clinic tool for the diagnosis and severity assessment
of a variety of retinal conditions.
Commercial Relationships: Lawrence Langley, None; Ben Davis,
None; Li Guo, None; Lisa Turner, None; Shereen Nizari, None; M
Francesca Cordeiro, None
Projection of the 3D OCT image of a rat retina with pattern bleach.
(a) dark-adapted image; (b) normalized differential image between
dark- and light-adapted states. The vertical strip is bleached before
the dark-adapted OCT image is taken.
Commercial Relationships: Shuliang Jiao, None; Tan Liu, None;
Byron L. Lam, None; Rong Wen, None
Support: 5R01EY019951-04

Multidisciplinary Ophthalmic Imaging Group 139 Clinical

Transcription

Similar documents

Practice Brochure - Arizona Retina Institute

Color Variability Analysis in Fundus Photography

Retinal Toxicity - Healthcare Professionals

Imaging Unlimited

the monaciano symposium - Fondo Studenti Italiani

LECTURE 9 Posterior Pole retina pathology_HTN_DM retinopathy

TRANSPORTATION

Bull`s Eye Maculopathy - SUNY Downstate Medical Center

Things about the Eye that every Internist should know COLD!!!

FACeTS FACeTS - A Snowmobile for George