Iterative Reconstruction Techniques (ASIR, MBIR, IRIS): New

Iterative Reconstruction Techniques (ASIR, MBIR, IRIS): New

Iterative Reconstruction
Techniques (ASIR, MBIR, IRIS):
New Opportunities to Lower Dose
and Improve Image Quality
Dushyant Sahani MD
Director of CT
Associate Professor of Radiology
MGH
email: [email protected]
HARVARD
MEDICAL SCHOOL
Disclosures
Research Grant Support from GE
Healthcare
Steps for Lowering Dose
DECREASE IN
• Tube current
• Tube potential
• Gantry rotation time *
• Scan length
• Overlap scanning
• Number of scan phases
INCREASE IN
• Pitch (Table speed) *
•Beam collimation
*Provided other scanning factors
are kept constant
MGH Webster Center for Radiation Dose Research and Education
Image Quality (IQ) and Dose
Optimum
Reduced
Low dose scan
Lower radiation dose reduced image quality
MGH Webster Center for Radiation Dose Research and Education
Low Dose CT Challenges:
Image Quality
Scan
Parameters
Diagnostic
Performance
Safety
(dose)
Image
Quality
NI-15
NI-15
NI-35
NI-35
Low Dose Imaging
• Standard Filtered back projection (FBP)
treats projection data as if coming form an
idealized system
• Efficient but require more dose for IQ
• Excessive image noise at low dose
• Artifacts in large patients and dense
structures
• Reduced CNR
How Does IR helps ?
Optimum
Reduced
IR removes the noise and return the image quality
MGH Webster Center for Radiation Dose Research and Education
Chest CT at 25% Dose
Low-Dose Imaging
FBP
In x-ray CT, lowering radiation
dosage increases noise
Fewer photon penetration to
detector
Iterative reconstruction
algorithms can increase image
signal-to-noise ratio
Can potentially produce same
diagnostic accuracy /
confidence as full-dose CT,
but at a fraction of the
radiation
Courtesy: H Pien, MGH
Iter. Recon
Iterative Recon – Low Dose
Phantom
Filtered Backprojection
Reconstructions
100% Dose
75% Dose
50% Dose
25% Dose
IRT
Courtesy: H Pien, MGH
Resolution Improvement
10 lp/cm
FBP46f(100%D)
FBP46f(50%D)
IRT(50%D)
Contrast window [500,2000]
Courtesy: H Pien, MGH
Standard FBP
Statistical IR (ASiR)
Fast Image
Data Space
IRIS
Fast Image
Data Space
Fast Image
Data Space
Basic
image
correction
Image data
recon
Exact image
correction
Compare
Raw data
recon
Raw data
recon
Basic
raw data
projection
Compare
Slow Raw
Data Space
Slow Raw
Data Space
Slow Raw
Data Space
FBP
Simple
VS.
Comprehensive
Simple
Fast
High noise
VS.
CPU
Better IQ
Low dose
Low noise
Full IR
Advanced
Few
Iterations
Ideal
System
CPU
Partial IR
Advanced
model
CPU
MGH Webster Center for Radiation Dose Research and Education
Optimal IQ
Low noise
Better resolution
Iterative Reconstruction
System Model
Raw Data
MBIR Image
Cost
Function
Data Statistics
Nonlinear
Smoothing
Optimization
(Recon)
• Optimal use of spatiallyvarying statistical model for
stable, low noise, and high
resolution IQ
Courtesy: Bob Becket, GE Healthcare
NO ASIR
30 ASIR
60 ASIR
10 ASIR
40 ASIR
70 ASIR
20 ASIR
50 ASIR
100 ASIR
Jan 10 2008
Crohn’s Disease
Dec 20 2008
23 yrs old with
Crohn’s Disease.
Baseline and FU CT
exam
40% ASIR
55% Dose
Reduction
Radiation Dose – 11.6mSv
Radiation Dose – 5.1 mSv
July 2 2007
Kidney Stone
Feb 26 2009
60%
ASIR
263 lbs
DLP 744
11.16mSv
81%
DLP 138
2.07mSv
NI = 28.7 kV= 120, 0.5s rotation, 1.375 pitch, 55mm/rot
CTDIVol = 9.72mGy DLP = 456.73 mGy-cm
FBP
FBP + 60% ASIR
MBIR
Coronal Reformat at 0.625mm thickness
FBP no ASIR
FBP w/ 50% ASIR
MBIR
NI=24.2, kV= 120, 0.5s rotation, 1.375 pitch, 55mm/rot
CTDIVol = 12.44mGy DLP = 609.26 mGy-cm
Summary
• Commercially available Iterative techniques (ASiR/IRIS/MBIR)
can improve image quality of low dose exams
• Dual benefits of improved Image Quality and dose for patients
with large body habitus
• Enough experience with ASiR and IRIS to justify their role in
clinical use
• Higher cost and available only for high end scanners (64-MDCT)
• Potential work flow limitations (slower data processing) with
MBIR/Veo