Artifacts in ECG-Synchronized MDCT Coronary Angiography

09_07_2138_Kroft.fm — 7/27/07
Kroft et al.
Artifacts in ECGSynchronized MDCT
Coronary Angiography
Cardiac Imaging • Review
Artifacts in ECG-Synchronized
MDCT Coronary Angiography
L. J. M. Kroft1
A. de Roos
J. Geleijns
Kroft LJM, de Roos A, Geleijns J
OBJECTIVE. In MDCT coronary angiography, image artifacts are the major cause of
false-positive and false-negative interpretations regarding the presence of coronary artery
stenoses. Hence, it is important that observers reporting these investigations are aware of the
potential presence of image artifacts and that these artifacts are recognized.
CONCLUSION. The article explores the technical causes for various artifacts in MDCT
coronary angiography imaging and clinical examples are given.
rom the first 4-MDCT feasibility
studies to the current clinically
applied 64-MDCT investigations, MDCT coronary angiography has evolved into a reliable 3D imaging
technique for detecting and excluding coronary artery stenoses with high accuracy
[1–8]; it is now considered an appropriate
imaging tool for detecting coronary artery
disease in certain clinical contexts [9]. However, 2D conventional invasive X-ray coronary angiography is still considered the standard of reference for evaluating the coronary
arteries because of its superior spatial and
temporal resolutions compared with MDCT.
Parts of the coronary arteries cannot be evaluated with MDCT because of image artifacts, and image artifacts are the major cause
of false-positive and false-negative interpretations. Hence, it is important that observers
reporting MDCT coronary artery angiography investigations are aware of the potential
presence of image artifacts and that these artifacts are recognized on the images.
In this article, the impact of artifacts on
MDCT coronary angiography will be explained. The causes of artifacts will be discussed in detail, with special attention to the
effect of coronary artery size and motion. Examples of artifacts will be shown that may
help the reader recognize these artifacts when
reporting MDCT coronary angiography.
F
Keywords: artifacts, cardiac imaging, computed
tomography, coronary angiography, CT, diagnostic imaging
DOI:10.2214/AJR.07.2138
Received December 11, 2006; accepted after revision
March 8, 2007.
1All authors:
Department of Radiology, C2S, Leiden
University Medical Center, Albinusdreef 2, 2333 ZA, Leiden,
The Netherlands. Address correspondence to L. J. M. Kroft
([email protected]).
CME
This article is available for CME credit. See www.arrs.org
for more information.
AJR 2007; 189:581–591
0361–803X/07/1893–581
© American Roentgen Ray Society
AJR:189, September 2007
Impact of Artifacts on MDCT
Coronary Angiography
MDCT coronary angiography image
quality and diagnostic performance have
greatly improved after recent technical developments. With 4-MDCT, 29% of the coronary arteries could not be evaluated because of artifacts [1]. With 16-MDCT,
22–29% of the coronary artery segments
could not be evaluated [10, 11]. One study
stated that if these segments that could not
be evaluated were excluded or considered
negative, 25% of patients with a significant
stenosis would have been missed [11]. With
64-MDCT, 3–11% of coronary artery segments still cannot be evaluated [8, 12–15].
Sensitivities and specificities for detecting significant (≥ 50%) coronary artery
stenoses based on segmental analysis with
64-MDCT (conventional X-ray coronary
angiography as the standard of reference)
have been found to be good to excellent, in
the range of 76–99% and 95–97%, respectively [2–8, 16]. However, these study outcomes are difficult to compare because the
study methods vary substantially—for example, in the selection of patients. Moreover, these study results should be interpreted with care because coronary artery
segments that could not be evaluated
(3–27%) were excluded from analysis beforehand [2–5, 7, 8].
Interestingly, of the coronary artery segments that were included in 64-MDCT studies, the accuracy for detecting stenoses depended highly on image artifacts. Falsepositive and false-negative interpretations
were attributed to image artifacts in 91% [6]
to 100% [5] of cases, where the major cause
was the presence of calcifications. Other authors support these findings [2–4, 6]. Less
581
09_07_2138_Kroft.fm — 7/27/07
Kroft et al.
TABLE 1: Artifacts and Causes, Explanations, and Measures to Avoid Them
Artifact or Error
Cause
Technical Explanation
Solution
Blooming (artifact)
Small high-contrast objects appear larger
than they are (e.g., caused by dense
calcifications, clips, and stents)
Not ideal transfer function
(point-spread function)
Smallest available focal spot, optimal
reconstruction filter
Blurring (artifact)
Averaging of real attenuation values in a
voxel
Volume averaging due to voxel Thinner slices, smaller field of view
size
Blurring (artifact)
Smearing out of real attenuation values due Volume averaging due to
to cardiac motion, respiratory motion, or
motion
postural motion
β-blockers, nitroglycerin agents,
sedatives, breath-holding instructions to
patients
Streaks (artifact)
Reconstruction artifact caused by, e.g.,
Beam hardening
lumen filled with high iodine concentration,
metal clips, or metallic implants
Saline flush to avoid iodine contrast
artifacts, use of less dense metal clips and
implants
Subtle discontinuities and missing
data (artifact)
Respiratory or postural motion, pitch factor No data for proper image
too high, or poor-quality ECG signal
reconstruction
Proper instruction of patient, optimal fixing
of ECG leads, and checking of signal
Stairstep (artifact)
Lack of proper ECG
Nonoptimal phase selection during
synchronization
prospective trigger acquisition or during
retrospective reconstruction, due to either
tachycardia or patient arrhythmia
Manual ECG editing, cone-beam CT
(e.g., 256-MDCT scanner), β-blockers
Windmill (artifact)
Characteristic of spiral acquisitions; more Subsequent acquisitions,
prominent for multislice acquisitions and at interpolations, as between
high pitch factors
different detector rows
Optimization of pitch factor might reduce
windmill artifact; windmill artifacts are
avoided in cases of prospective triggered
acquisitions and cone-beam CT
(e.g., 256-MDCT scanner)
Instruct operator; update protocol; use
Poor contrast in lumen of coronary Technical error in contrast administration: Poor contrast enhancement
and resulting low contrast-to- high-density contrast agent, automatic
arteries (error)
extravasation (operator), volume and
(sure) start with accurate scanning delay,
speed (operator), timing (operator); patient noise ratio
higher mA and kV settings; instruct patient
factors, technical limitations: Valsalva
maneuver, heart failure, obesity
Incomplete coverage (error)
Technical error by operator
frequent causes were motion artifacts [4, 5]
and obesity resulting in a poor contrast-tonoise ratio [4]. It can be concluded that artifacts are the cause of coronary arteries that
cannot be evaluated and for false-positive
and false-negative diagnoses as well.
MDCT and Artifacts
Artifact Definition
In CT, the term “image artifact” can be defined as any discrepancy between the reconstructed Hounsfield values in the image and
the true attenuation coefficients of the object
in such a way that these discrepancies are
clinically significant or relevant as judged by
the radiologist [17]. In the definition, it is assumed that the radiologist will recognize relevant artifacts when they are present.
Technical Considerations for Coronary
Artery Imaging and Artifacts
Coronary artery MDCT is technically complex and requires high spatial resolution, high
temporal resolution, good low-contrast resolution, intravascular contrast enhancement, and a
582
Incorrect planning of scanning Instructions to operator
range
short scanning time. The key acquisition parameters in cardiac MDCT are section thickness, the rotation time of the X-ray tube, and the
pitch factor. The acquisition section thickness is
measured along the z-axis and determines the
minimal voxel height in a reconstructed image.
The rotation time (in milliseconds) is the time
needed for a 360° rotation of the X-ray tube.
The pitch factor is the ratio of patient displacement along the z-axis direction per tube rotation
divided by the total thickness of all simultaneously acquired sections [18].
Basically, almost all MDCT image artifacts can be explained by limitations relating
to spatial resolution, temporal resolution,
noise (Table 1), and the reconstruction algorithms used. In the images, artifacts are
mainly observed as blurring, blooming,
streaks, missing data, discontinuities, and
poor contrast enhancement. Artifacts may be
grouped in technical- (physics-based, scanner-based, and reconstruction-based), operator-, and patient-related causes [19]. Defining
the different artifact categories seems rather
arbitrary. For example, poor patient instruc-
tion that results in breathing artifacts may be
categorized as an operator-dependent artifact,
whereas breathing artifacts that occur despite
adequate breath-holding instructions may be
considered patient-dependent.
Spatial Resolution, Temporal
Resolution, Noise, and
Reconstruction Algorithm
Spatial Resolution
Spatial resolution, the ability to visualize
small structures in the scanned volume, must
be considered in three dimensions. A voxel
is the volume element that is represented by
a 2D pixel in the axial xy-plane. The third dimension is the z-axis and the corresponding
voxel height. A reconstructed field of view
of 200 mm and a characteristic 512 × 512
pixel matrix result in a pixel size of 0.4 × 0.4
mm2 in the axial xy-plane. These values are
typical for single-detector helical CT acquisitions and for the current MDCT scanners.
Reconstructions with a smaller field of view
can substantially decrease the pixel size in
the axial plane.
AJR:189, September 2007
09_07_2138_Kroft.fm — 7/27/07
Artifacts in ECG-Synchronized MDCT Coronary Angiography
> FWHM
< FWHM
FWHM
FWHM
A
B
Fig. 1—Principle of full width at half maximum (FWHM) of response of very small object for describing spatial resolution that can be achieved.
A and B, Visualization of two ideal points separated by distance of less than one FWHM (A) and separated more than one FWHM (B). Response of ideal point is represented
by gray area; composite response of two ideal points is represented by curving black line. At separation distance of less than one FWHM, two points cannot be distinguished
separately; at distance of more than one FWHM, two points can be observed individually. Note that this criterion assumes static condition, or, in other words, that no motion
artifacts are present.
The great improvement in spatial resolution with the current MDCT scanners is due
to the feasibility of acquiring volumes with
thinner sections, which is especially important for reduction of the partial volume effect. With 4-MDCT, 4 × 1 mm or 2 mm collimation scans were obtained in 20- to 45second breath-holds [1, 20–22]. With 16MDCT, 16 × 0.5 mm or 0.75 mm collimation scans were obtained in 16- to 30-second
breath-holds [23–25]. With 64-slice and 64row MDCT scanners, 64 × 0.6 mm and
64 × 0.5 mm collimations are achieved
within 11- to 15-second breath-holds [2, 5,
6]. Thinner collimation causes smaller
voxel height, and shorter breath-holds allow
more patients to hold their breath during image acquisition. Also, the scanned volume is
obtained in substantially fewer heart beats,
thereby decreasing the amount of image artifacts due to variations between heart beats.
However, voxel size alone is not sufficient for describing the spatial resolution
that can be achieved with CT scanners.
Voxel size does not take into account the inherent physical limitations of the CT scanner such as focal spot size, detector size, and
geometry (focus-to-axis of rotation distance
and focus-to-detector distance) that result in
geometric unsharpness [26]. The unambiguous and clearly defined quantity that physicists use for describing spatial resolution is
the full width at half maximum (FWHM) of
the response of a very small object. In CT, a
AJR:189, September 2007
tungsten bead with a submillimeter diameter
is often used as the small object for which
the FWHM is determined. Once the FWHM
is established for a scanner, one knows at
what separation distance two points can be
distinguished separately and at what separation distance two points are perceived as one
point. If two ideal points are separated just
a distance FWHM apart (or more), there is a
fair chance that they will be separated in the
image. If the points are separated less than a
distance FWHM, then the points will be visualized as one point (Fig. 1). Typical values for the FWHM in the axial plane are
0.5–0.7 mm; the FWHM along the z-axis is
slightly worse, with typical values between
0.7 and 1.0 mm [27]. This means that structures with dimensions smaller than the
FWHM of the point-spread function will be
severely distorted in the reconstructed images, even if a small reconstructed field of
view with correspondingly small voxels is
being used. In clinical applications of
MDCT coronary angiography, this aspect
should be taken into account—for example,
when coronary artery plaque (characterization) is considered and for imaging of small
coronary artery vessels.
Partial Volume Effect
Partial volume effect or artifact is caused
by limited spatial resolution and is the result
of averaging the attenuation coefficient in a
voxel that is heterogeneous in composition,
where the average numeric value (in Hounsfield units) is assigned to the corresponding pixel. Partial volume effect is especially
present at borders of two tissues or structures with a large difference in Hounsfield
units, particularly at the margin of the coronary artery lumen and calcified plaque. The
larger the voxels (i.e., pixel size and reconstructed image thickness), the larger the partial volume effect [18]. Partial volume artifacts are most often due to the presence of
calcifications and are a major concern in
MDCT coronary angiography because they
cause false-positive and false-negative interpretations in coronary arteries that could
otherwise be evaluated [2–6]. Partial volume artifacts, including blurring and
blooming, are best avoided by using thin
collimation and a small reconstructed field
of view [19].
Temporal Resolution
Temporal resolution is the ability to resolve
fast-moving objects in the displayed CT image
[18]. Temporal resolution remains the major
challenge in MDCT coronary angiography. Limitations in temporal resolution are strongly related to coronary artery size and motion. Motion
in general causes degradation of contrast and spatial resolution and introduces artifacts [28]. Cardiac motion presents as blurring and is the major
reason for nondiagnostic coronary artery image
quality [11–13]. The three major approaches to
limit cardiac motion artifacts were already postu-
583
09_07_2138_Kroft.fm — 7/27/07
Kroft et al.
A
B
Fig. 2—59-year-old man imaged for suspected
coronary artery disease. Stairstep artifact due to
premature atrial contraction with extra systole,
followed by compensatory long R-R interval (between
sixth and seventh R-R peaks)
A–D, Note that premature beat is approximately in
middle of acquisition (A), which is also true in images
(B–D). Note stairstep in right coronary artery (RCA)
(arrows) at 3D reconstructions and central luminal line
projections (B) and in two long-axis perpendicular
curved multiplanar reconstructions (C, D). In these
perpendicular curved multiplanar reconstructions,
coronary artery is usually more affected in one direction
than in other. Step had virtually no effect on left anterior
descending coronary artery (LAD in B). Mean heart rate
was 59 beats per minute. R-R interval during acquisition
varied between 644 and 1,281 milliseconds.
C
D
lated in the 1970s by Harell et al. [29]: reducing
the data collection period by faster rotating X-ray
tubes, synchronizing CT data acquisition with the
cardiac cycle (prospective gating), and reconstructions synchronized to the ECG cycle (retrospective reconstructions) after data acquisition.
These measures are currently routinely applied in
MDCT coronary angiography.
Irregular heart rates cause stairstep artifacts due to phase misregistration, in which
images are not reconstructed at exactly the
same phase of the heart cycle [30] (Fig. 2). Irregular heart rates—for example, premature
atrial contraction—can also cause blurring of
a coronary artery segment (Fig. 3). ECG editing may reduce these artifacts.
quality than those segments larger than 2.0
mm at 64-MDCT [14].
The coronary arteries move substantially
during the cardiac cycle with considerable intra- and interpatient variation regarding motion
patterns and ranges [32]. The right coronary artery has the greatest velocity and range [33,
34]. Transverse (axial xy-plane) displacement
is the major part of motion and ranges between
6 and 42 mm for the right and between 3 and
20 mm for the left coronary artery [32]. Motion
velocity increases with heart rate [33, 34], but
motion range does not [32]. The right coronary
artery is affected most by motion artifacts [12].
Excessive coronary artery motion of an inplane distance greater than its diameter corresponds to visible motion on transverse images
[33]. This amount of “own diameter motion”
seems a rather comfortable criterion, particularly when the limitations in temporal resolution affect CT of small structures (small-diameter coronary artery segments). With the
Coronary Artery Size and Motion
Mean coronary artery lumen diameters
vary substantially from their proximal to their
distal parts [31]. More than half the segments
are smaller than 2.0 mm and have less image
584
current MDCT scanners, coronary artery motion exceeds the velocities needed for motionfree imaging when variations in cardiac motion during the cardiac cycle are not taken into
account. However, motion velocity and speed
change during the cardiac cycle, and we must
use the “rest phase” for imaging with the fewest motion artifacts (Fig. 4). But even then,
the cardiac rest period, defined as the time
with a displacement of the coronary artery of
less than 1 mm, has a mean duration of 120
milliseconds but ranges from 66 to 333 milliseconds among patients [32]. Consequently,
the temporal resolution of 165 milliseconds
that is currently achieved with the fastest 330millisecond rotation times when using halfscan reconstructions is longer than the mean
rest period of 120 milliseconds and is not fast
enough to image the general patient free of
motion artifacts, notwithstanding optimal
phase selection for image reconstruction in
the cardiac cycle.
AJR:189, September 2007
09_07_2138_Kroft.fm — 7/27/07
Artifacts in ECG-Synchronized MDCT Coronary Angiography
Fig. 3—54-year-old woman with suspected coronary artery disease. Image shows blurring due to motion caused
by premature atrial contraction. At short R-R interval, rest phase was too short for motion-free imaging of coronary
artery segment that presumably had large motion range at this time, causing blurring. This segment of right
coronary artery is frequently affected by motion artifacts. Mean heart rate was 66 beats per minute. R-R interval
during acquisition varied between 641 and 1,194 milliseconds.
Fig. 4—51-year-old woman with suspected coronary artery disease. Image shows motion range for right coronary
artery (RCA) during cardiac cycle. Image reconstructions were performed at 0%, 40%, and 80% of R-R interval and
show identical orientation of 3D images in upper row and identical levels of images in middle row. In lower row,
level that best displayed origin of RCA is displayed. Note large amount of motion of RCA during cardiac cycle. Note
that RCA is displayed sharply at 80% of R-R interval, but not at 0% and 40% time phases. Mean heart rate was 52
beats per minute. R-R interval during acquisition varied between 1,095 and 1,189 milliseconds.
AJR:189, September 2007
With heart rates of less than 65 beats per
minute (bpm), the best image quality is predominantly in diastole, whereas in heart rates
exceeding 75 bpm, the best image quality
shifts to systole in most cases [1, 13, 35]. Increasing heart rates mainly affect and significantly shorten the rest period at diastole [1, 32].
Choosing the optimal time point for reconstruction becomes more crucial for preserving
image quality with higher heart rates [34].
With 64-MDCT and at heart rates less than 65
bpm, diagnostic image quality can be obtained
for all coronary arteries at a single reconstruction interval at mid diastole [15]. When higher
heart rates are included, a minority of patients
(7%) require additional reconstructions at late
systole for optimal visualization of the right
coronary artery [12].
To overcome limitations in temporal resolution and motion artifacts, increased gantry rotation speed is favored. However, a rotation
time of less than 200 milliseconds to provide a
temporal resolution of less than 100 milliseconds, regardless of cardiac frequency (halfscan reconstruction), would already result in
an increase in mechanical G-forces that is beyond mechanical engineering limits [36].
Other strategies are as follows:
Lowering the cardiac frequency by using
β-blockers or dilating the coronary arteries
by using nitroglycerin—Coronary artery image quality is inversely dependent on heart rate
[1, 4, 12, 20, 21, 37]. Administering β-blockers
for coronary artery MDCT is widely used for
reducing the heart rate, preferably to a frequency of less than 60–65 bpm [2–5, 7, 8, 38].
Despite these publications, it has recently been
found that the effect of heart rate on image
quality is limited and mainly affects the visualization of the left circumflex coronary artery.
Instead, the variability in heart rate during acquisition was found to be important. The stabilizing effect of β-blockers was shown to be the
major determinant that resulted in superior image quality in patients receiving β-blockers as
compared with those who did not [39]. Some
authors use vasodilating medication as well for
maximizing coronary artery size [40, 41]. In
MDCT coronary angiography, the use of nitroglycerin has been found to increase proximal
coronary artery diameters by 12–21% [41].
However, the added value on diagnostic accuracy is not clear yet.
Using segmental reconstruction instead of
half-scan reconstruction—The single-segment
approach (i.e., half-scan reconstruction) and
the two or more segment approach (i.e., multisegment reconstruction) are the two image
585
09_07_2138_Kroft.fm — 7/27/07
Kroft et al.
prospective triggering) minimizes the
breath-holding time, allowing almost all
patients to breath-hold during image acquisition. Moreover, multisegment reconstruction—for example, two-segment reconstruction—may potentially be used to
improve the acquisition time.
A
C
B
D
Fig. 5—69-year-old woman with suspected coronary artery disease. Images show poor contrast enhancement.
Contrast timing was good because coronary arteries were already enhancing.
A–D, Note poor enhancement of left ventricle (LV), which should be brightly enhanced (B) (compare with Fig. 4). Also
note stent in circumflex coronary artery (A and C), where artery is moderately enhanced. Patient performed Valsalva
maneuver during image acquisition that is recognized by contrast column with convex shape toward superior vena
cava (SVC on coronal image, D), whereas saline flush should be running through at this time point. High intrathoracic
pressure during Valsalva maneuver hampers inflow in right atrium and causes poor contrast enhancement. Mean
heart rate was 77 beats per minute. R-R interval during acquisition varied between 776 and 789 milliseconds.
reconstruction algorithms used for low and
higher heart rates, respectively. With halfscan reconstruction, data obtained from a single 180° gantry rotation are used for image reconstruction [36]. With multisegment reconstruction, data of two or more successive
cardiac cycles are combined that cover, as
separate segments, 180° acquisition. Multisegment reconstruction algorithms require a
stable and predictable heart rate during image
acquisition. The temporal resolution is improved by a factor 2n (n = number of cycles
and segments) of the rotation time [36]. For
64-MDCT with 330 milliseconds rotation
time, it was found that image quality achieved
with two-segment reconstruction was not significantly improved compared with half-scan
reconstruction for heart rates exceeding 65
bmp, although in 65% of patients the best
overall image quality was achieved by twosegment reconstruction [13].
586
Dual or multiple-source CT scanners in
combination with half-scan reconstruction algorithms—A new technology, the dualsource MDCT technique has recently been
introduced with two tubes mounted at an
angle of 90°, thus improving the temporal
resolution by a factor 2. A rotation time of
330 milliseconds results in a temporal resolution of 83 milliseconds by half-scan reconstruction; this is achieved independently of heart rate [42]. Development of
scanners with more than two source/detector combinations may further improve temporal resolution.
Faster scanning with 256-MDCT volumetric scanners—With these next generation
volumetric MDCT scanners [43, 44], the
entire heart can be covered in only one half
rotation. This is expected to eliminate interbeat (stairstep) artifacts, and a short scanning time of 175 milliseconds (half-scan,
Respiratory Motion
Breath-holding exercises and instructions
to the patients are particularly important in
avoiding motion artifacts by breathing and
postural motions that cause blurring. With
the current 64-MDCT scanners, most patients can breath-hold for the 10- to 13-second scanning duration. We instruct the patients to hold their breath after breathing in,
at approximately three quarters of their maximum, and to lie still without pressing in order to avoid a Valsalva maneuver that may
result not only in breathing artifacts but also
in poor contrast enhancement (Fig. 5). Respiratory motion is well recognized at the
lung window setting.
Noise
CT noise (quantum mottle) is determined
primarily by the number of photons used to
make an image. The quantum mottle fraction decreases as the number of photons increases. CT noise is generally reduced by
increasing the kVp, mA, or scanning time.
CT noise is also reduced by increasing
voxel size (i.e., by decreasing the matrix
size), increasing reconstructed field of
view, by increasing section thickness [45],
or by image stacking.
Contrast is the difference in Hounsfield
values between tissues and tends to increase
as kVp decreases but is less affected by mA
or scanning time. CT contrast can be improved by administering an iodinated contrast agent. The displayed image contrast is
primarily determined by the CT windowwidth and window-level settings [45].
Larger chest sizes (obesity) are associated
with a higher level of image noise that negatively affects the quality of MDCT coronary angiograms [12]. Inadequate contrast administration (e.g., inadequate volume, injection speed,
or timing), inadequate selection of the field of
view or region-of-interest placement for bolus
tracking, or inadequate breath-holds can result
in low contrast-to-noise ratios, resulting in
poorly visualized coronary arteries. Contrast
media with higher iodine concentrations provide substantially higher attenuation values in
the coronary arteries [46], although the added
AJR:189, September 2007
09_07_2138_Kroft.fm — 7/27/07
Artifacts in ECG-Synchronized MDCT Coronary Angiography
A
B
C
D
Fig. 6—60-year-old woman with suspected coronary artery disease. Geometric distortion due to spiral acquisition,
where black “shadow” or “rod” artifact next to contrast-filled right coronary artery is due to miscalculation by
reconstruction algorithm. During spiral acquisition, position registered by each view shifts.
A–D, Miscalculation may cause hypodense artifacts (arrows) that rotate around high-density contrast-filled
coronary artery. Note change in artifact position from A to C that is also observed on corresponding levels at
coronal reconstruction (D). Mean heart rate was 66 beats per minute. R-R interval during acquisition varied
between 916 and 977 milliseconds.
value of these higher-iodine-concentration media on diagnostic accuracy in assessing coronary artery disease has not yet been established.
Artifacts on the right coronary artery due to
high contrast density in the right atrium can be
effectively reduced using a saline flush. However, the diagnostic accuracy in detecting coronary artery stenoses has not been found to be
substantially different between uni- or biphasic
contrast protocols with or without a saline flush
[47]. Probably, the contrast dose and injection
speed should be balanced for optimal enhancement. Testing the IV access before contrast administration is advisable and is best performed
just before coronary MDCT by flushing a saline bolus with the patient’s arms up in the same
position as during coronary artery scanning.
AJR:189, September 2007
Reconstruction Algorithm
Spiral acquisitions may cause artifacts due
to suboptimal (large) pitch, which is due to table movement during data acquisition that
causes subsequent projections to be acquired
from slightly different parts of the object. Resulting inconsistency in the data causes artifacts that increase with pitch. General CT artifacts due to spiral scanning in relation to
pitch are cone and rod artifacts [48]. With the
current generation of MDCT scanners, cone
artifacts are avoided by various dedicated
MDCT reconstruction approaches that account for cone-beam geometry [36]. Rod artifacts [48]—also referred to as “windmill” artifacts [36]—are caused by the spiralinterpolation process of high-contrast struc-
tures that are obliquely oriented along the zaxis scanning plane. During scanning, the position registered by each view shifts; the size
of the shift depends on the structure’s angulation and the table increment. Rod artifacts occur as hypodense (or hyperdense) structures
around structures with high (or low) density.
These helical CT artifacts are often observed
around the ribs, where hypodense artifacts rotate around dense ribs and are seen when
scrolling through a stack of images [36, 42].
Because of the shape, the position, and the
high contrast density of the coronary arteries,
rod artifacts may occur in coronary MDCT angiography as well (Fig. 6), and these artifacts
may hamper coronary artery evaluation. These
artifacts should be recognized and not be confused with noncalcified coronary plaque.
To reduce spiral interpolation artifacts,
narrow collimation that improves the z-axis
resolution should be used [36]. One may expect these spiral CT artifacts to disappear
with the introduction of newer generation
(256-row) scanners in which acquisition of
the entire cardiac volume will be obtained
without table movement.
Beam-hardening artifacts are caused by
the polychromatic nature of the X-ray beam.
As the lower energy photons are preferentially absorbed, the beam becomes more
penetrating, which results in lower computed attenuation coefficient Hounsfield
values. Beam-hardening artifacts are most
prominent at high-contrast interfaces [45].
In cardiac imaging, high-density contrast
agent injection can manifest as beam-hardening artifacts by causing dark bands between dense objects in the image [17].
Metal objects can cause complicated artifacts, such as beam hardening and partial
volume, that are worsened with object motion, and methods to overcome metal-induced artifacts are particularly difficult to
design [17]. In MDCT coronary angiography, these artifacts may occur with stents,
pacemakers, or surgical clips. If the density
of highly attenuating metal objects is beyond the normal range that can be handled
by the computer, severe streaking artifacts
occur [19] (Fig. 7). Sharp high-resolution
kernels may be used for improved stent–lumen visualization [49], although sharp kernel filters result in higher image noise and
artifacts causing lower in-stent attenuation
values [50]. The diagnostic effect of highresolution kernels on accuracy in evaluating stent patency in patients has not been
investigated yet.
587
09_07_2138_Kroft.fm — 7/27/07
Kroft et al.
Fig. 7—Two patients
with suspected coronary
artery disease.
A and B, 34-year-old man
with pacemaker lead in
right atrium (B) that
causes subtle artifacts
visible at right ventricular
surface in 3D view
(arrows, A) and through
right coronary artery
central luminal line
reconstruction (arrow,
B). Mean heart rate was
78 beats per minute. R-R
interval during
acquisition varied
between 759 and 790
milliseconds.
C and D, 57-year-old man
after bypass surgery with
metallic sternal wires (C).
Severe high-density
surgical clip artifacts
hamper arterial lumen
evaluation at course of
left internal mammary
artery, which was used
for bypassing left
anterior descending
coronary artery (D).
Surgical clips were used
for occluding side
branches of left internal
mammary artery. Mean
heart rate was 74 beats
per minute. R-R interval
during acquisition varied
between 760 and 835
milliseconds.
Reconstruction and Postprocessing Errors
Errors during reconstruction can occur because of inadequate or insufficient phase selection for coronary artery evaluation. Multiple-phase reconstructions may be needed for
visualizing all coronary artery segments with
diagnostic quality (Fig. 8). At the workstation, automatic tools for segmentation may
produce errors (Fig. 9). It is important to always review the nonpostprocessed source images to confirm findings found at the computer-assisted reconstructions.
Diagnostic Interpretation Errors
For 64-MDCT, false-positive and falsenegative coronary artery interpretations
588
A
B
C
D
have been explicitly explained by obvious
technical limitations—that is, image artifacts due to calcifications, motion, and obesity [2–6]. This was also the case for 4MDCT [21] and 16-MDCT [10, 11] coronary angiography. Therefore, missing coronary artery abnormalities seems to be most
commonly related to artifacts and less related to lack of diagnostic perception in the
case of well-trained observers.
Technical errors that can be avoided are
often related to patient handling and postprocessing handling. Suggestions for avoiding MDCT coronary angiography artifacts
are presented in Table 1. Global measures
are the following:
Patient preparation—Lower the heart rate
in patients with heart rates exceeding 65
bpm by using β-blockers (if allowed). Obtain a good ECG signal. Prepare contrast injection and prepare the patient with breathholding instructions.
Acquisition—Take care for adequate timing of contrast injection, and use an adequate
dose and injection speed. Provide good
breath-holding instructions.
Postprocessing—Choose the best phase for
coronary artery reconstruction. Review the
source images to confirm findings at advanced workstation reconstructions.
Diagnostic interpretation—Recognize artifacts and report diagnostic limitations.
AJR:189, September 2007
09_07_2138_Kroft.fm — 7/27/07
Artifacts in ECG-Synchronized MDCT Coronary Angiography
A
B
C
D
Fig. 8—77-year-old woman with suspected coronary
artery disease.
A–D, Curved multiplanar reconstruction of right
coronary artery in two perpendicular longitudinal
directions. At time point with least motion at 900
milliseconds (at 76% of R-R interval, A and B), right
coronary artery is sharply delineated in proximal part
but appears interrupted halfway (arrow), whereas
further coronary artery segment appears blurred.
Severe stenosis cannot be excluded at this time.
Additional reconstruction at 500 milliseconds (at 42%
of R-R interval, C and D) is of moderate but diagnostic
quality and shows that suspected right coronary artery
segment is actually open. Mean heart rate was 51
beats per minute. R-R interval during acquisition varied
between 1,155 and 1,198 milliseconds.
B
A
Fig. 9—51-year-old man with suspected coronary artery disease.
A–C, Automatic segmentation reconstruction artifact with interruption and apparent stenosis of right coronary artery
(RCA) at crux where it diverges into RCA continuation in atrioventricular groove and in posterior descending branch
(PD). Point of interruption is where central luminal line (dotted line, A) has lost its way at crux and is not in center of
coronary artery (A). This is easily repaired by manually replacing erroneous point (A) correctly in central lumen at
curved multiplanar reconstruction. After replacement, artery is continuous; compare B and C. Mean heart rate was
52 beats per minute. R-R interval during acquisition varied between 1,124 and 1,172 milliseconds.
C
AJR:189, September 2007
589
09_07_2138_Kroft.fm — 7/27/07
Kroft et al.
Conclusion
The main artifacts that hamper MDCT coronary angiography image interpretation are
motion artifacts that cause blurring and incorrect diagnoses due to coronary artery calcifications. This article has explored artifact causes
and provided examples. Recognizing artifacts
is important in the diagnostic process.
Acknowledgment
We thank Raoul M.S. Joemai for producing
Figure 1.
10.
11.
References
1. Giesler T, Baum U, Ropers D, et al. Noninvasive
visualization of coronary arteries using contrast-enhanced multidetector CT: influence of heart rate on
image quality and stenosis detection. AJR 2002;
179:911–916
2. Mollet NR, Cademartiri F, van Mieghem CAG, et
al. High-resolution spiral computed tomography
coronary angiography in patients referred for diagnostic conventional coronary angiography. Circulation 2005; 112:2318–2323
3. Leber AW, Knez A, von Ziegler F, et al. Quantification of obstructive and nonobstructive coronary
lesions by 64-slice computed tomography: a comparative study with quantitative coronary angiography and intravascular ultrasound. J Am Coll Cardiol
2005; 46:147–154
4. Raff GL, Gallagher MJ, O’Neill WW, Goldstein JA.
Diagnostic accuracy of noninvasive coronary angiography using 64-slice spiral computed tomography. J Am Coll Cardiol 2005; 46:552–557
5. Pugliese F, Mollet NRA, Runza G, et al. Diagnostic
accuracy of non-invasive 64-slice CT coronary angiography in patients with stable angina pectoris.
Eur Radiol 2006; 16:575–582
6. Leschka S, Alkadhi H, Plass A, et al. Accuracy
of MSCT coronary angiography with 64-slice
technology: first experience. Eur Heart J 2005;
26:1482–1487
7. Ropers D, Rixe J, Anders K, et al. Usefulness of
multidetector row spiral computed tomography
with 64- x 0.6-mm collimation and 330-ms rotation for the non-invasive detection of significant
coronary artery stenoses. Am J Cardiol 2006;
97:343–348
8. Nikolaou K, Knez A, Rist C, et al. Accuracy of 64MDCT in the diagnosis of ischemic heart disease.
AJR 2006; 187:111–117
9. Hendel RC, Patel MR, Kramer CM, Poon M.
ACCF/ ARC/ SCCT/ SCMR/ ASNC/ NASCI/
SCAI/ SIR 2006 appropriateness criteria for cardiac
computed tomography and cardiac magnetic resonance imaging: a report of the American College of
Cardiology Foundation Quality Strategic Direc-
590
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
tions Committee Appropriateness Criteria Working
Group, American College of Radiology, Society of
Cardiovascular Computed Tomography, Society
for Cardiovascular Magnetic Resonance, American
Society of Nuclear Cardiology, North American
Society for Cardiac Imaging, Society for Cardiovascular Angiography and Interventions, and Society of Interventional Radiology. J Am Coll Cardiol
2006; 48:1475–1497
Heuschmid M, Kuettner A, Schroeder S, et al. ECGgated 16-MDCT of the coronary arteries: assessment of image quality and accuracy in detecting
stenosis. AJR 2005; 184:1413–1419
Garcia MJ, Lessick J, Hoffmann MHK. Accuracy
of 16-row multidetector computed tomography for
the assessment of coronary artery stenosis. JAMA
2006; 296:403–411
Ferencik M, Nomura CH, Maurovich-Horvat P, et
al. Quantitative parameters of image quality in 64slice computed tomography angiography of the coronary arteries. Eur J Radiol 2006; 57:373–379
Wintersperger BJ, Nikolaou K, von Ziegler F, et al.
Image quality, motion artifacts, and reconstruction
timing of 64-slice coronary computed tomography
angiography with 0.33-second rotation speed. Invest Radiol 2006; 41:436–442
Pannu HK, Jacobs JE, Lai S, Fishman EK. Coronary
CT angiography with 64-MDCT: assessment of
vessel visibility. AJR 2006; 187:119–126
Leshka S, Husmann L, Desbiolles LM, et al. Optimal image reconstruction intervals for non-invasive
coronary angiography with 64-slice CT. Eur Radiol
2006; 16:1964–1972
Mühlenbruch G, Seyfarth T, Soo CS, Pregalathan
N, Mahnken AH. Diagnostic value of 64-slice
multi-detector row cardiac CTA in symptomatic patients. Eur Radiol 2007; 17:603–609
Hsieh J. Image artifacts: appearances, causes, and
corrections. In: Computed tomography: principles,
design, artifacts, and recent advances. Bellingham,
WA: SPIE Press, 2003:167–240
Geleijns J, Kroft LJM, Bax JJ, Lamb HJ, de Roos
A. Techniques for cardiovascular computed tomography. In: Higgins CB, de Roos A, eds. MRI and CT
of the cardiovascular system, 2nd ed. Philadelphia,
PA: Lippincott Williams & Wilkins, 2006:37–52
Barrett J, Keat N. Artifacts in CT: recognition and
avoidance. RadioGraphics 2004; 24:1679–1691
Hong C, Becker CR, Huber A, et al. ECG-gated reconstructed multi-detector row CT coronary angiography: effect of varying trigger delay on image
quality. Radiology 2001; 220:712–717
Nieman K, Rensing BJ, van Geuns RJM, et al. Noninvasive coronary angiography with multislice spiral computed tomography: impact of heart rate.
Heart 2002; 88:470–474
Dirksen MS, Jukema JW, Bax JJ, at al. Cardiac
multidetector-row computed tomography in pa-
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
tients with unstable angina. Am J Cardiol 2005;
95:457–461
Dewey M, Laule M, Krug L, et al. Multisegment
and half-scan reconstruction of 16-slice computed
tomography for detection of coronary artery
stenoses. Invest Radiol 2004; 39:223–229
Kantarci M, Ceviz N, Durur I, et al. Effect of the reconstruction window obtained at isovolumetric relaxation period on the image quality in electrocardiographic-gated 16-multidetector-row computed
tomography coronary angiography studies. J Comput Assist Tomogr 2006; 30:258–261
Hoffmann MHK, Lessick J, Manzke R, et al. Automatic determination of minimal cardiac motion
phases for computed tomography imaging: initial
experience. Eur Radiol 2006; 16:365–373
Hsieh J. Helical or spiral CT. In: Computed tomography: principles, design, artifacts, and recent advances. Bellingham, WA: SPIE Press,
2003:265–305
Hsieh J. Important performance parameters. In:
Computed tomography: principles, design, artifacts, and recent advances. Bellingham, WA: SPIE
Press, 2003:113–145
Alfidi RJ, Mac Intyre WJ, Haaga JR. The effects
of biological motion on CT resolution. AJR 1976;
127:11–15
Harell GS, Guthaner DF, Breiman RS, et al. Stopaction cardiac computed tomography. Radiology
1977; 123:515–517
Hsieh J. Advanced CT applications. In: Computed
tomography: principles, design, artifacts, and recent advances. Bellingham, WA: SPIE Press,
2003:341–382
Dodge JT, Brown BG, Bolson EL, Dodge HT. Lumen diameter of normal human coronary arteries:
influence of age, sex, anatomic variation, and left
ventricular hypertrophy or dilation. Circulation
1992; 86:232–246
Wang Y, Vidan E, Bergman GW. Cardiac motion of
coronary arteries: variability in the rest period and
implications for coronary MR angiography. Radiology 1999; 213:751–758
Lu B, Mao SS, Zhuang N, et al. Coronary artery motion during the cardiac cycle and optimal ECG triggering for coronary artery imaging. Invest Radiol
2001; 36:250–256
Achenbach S, Ropers D, Holle J, Muschiol G,
Daniel WG, Moshage W. In-plane coronary arterial
motion velocity: measurement with electron-beam
CT. Radiology 2000; 216:457–463
Herzog C, Abolmaali N, Balzer JO, et al. Heartrate-adapted image reconstruction in multidetectorrow cardiac CT: influence of physiological and
technical prerequisite on image quality. Eur Radiol
2002; 12:2670–2678
Flohr TG, Schaller S, Stierstorfer K, Bruder H,
Ohnesorge BM, Schoepf UJ. Multi-detector row
AJR:189, September 2007
09_07_2138_Kroft.fm — 7/27/07
Artifacts in ECG-Synchronized MDCT Coronary Angiography
37.
38.
39.
40.
41.
CT systems and image-reconstruction techniques.
Radiology 2005; 235:756–773
Schroeder S, Kopp AF, Kuettner A, et al. Influence
of heart rate on vessel visibility in noninvasive coronary angiography using new multislice computed
tomography: experience in 94 patients. J Clin Imaging 2002; 26:106–111
Lawler LP, Pannu HK, Fishman EK. MDCT evaluation of the coronary arteries, 2004: how we do
it—data acquisition, postprocessing, and interpretation. AJR 2005; 184:1402–1412
Leschka S, Wildermuth S, Boehm T, et al. Noninvasive coronary angiography with 64-section
CT: effect of average heart rate and heart rate
variability on image quality. Radiology 2006;
241:378–385
Ferencik M, Nomura CH, Maurovich-Horvat P, et
al. Quantitative parameters of image quality in 64slice computed tomography angiography of the coronary arteries. Eur J Radiol 2006; 57:373–379
Dewey M, Hoffmann H, Hamm B. Multislice CT
42.
43.
44.
45.
coronary angiography: effect of sublingual nitroglycerin on the diameter of the coronary arteries.
Rofo 2006; 178:600–604
Achenbach S, Ropers D, Kuetter A, et al. Contrastenhanced coronary artery visualization by dualsource computed tomography: initial experience.
Eur J Radiol 2006; 57:331–335
Funabashi N, Yoshida K, Tadokoro H, et al. Three
dimensional segmented myocardial perfusion images by selective intracoronary injection of contrast
using 256 slice cone beam computed tomography.
Heart 2005; 91:1349–1351
Mori S, Kondo C, Suzuki N, Hattori A, Kusakabe M, Endo M. Volumetric coronary angiography using the 256-detector row computed tomography scanner: comparison in vivo and in
vitro with porcine models. Acta Radiol 2006;
47:186–191
Huda W, Slone RM. Computers and computed tomography (CT). In: Nieginski EA, Dinkel AG, eds.
Review of radiologic physics. Philadelphia, PA:
Lippincott Williams & Wilkins, 1995:93–110
46. Cademartiri F, Mollet NR, van der Lugt A, et al. Intravenous contrast material administration at helical
16-detector row CT coronary angiography: effect of
iodine concentration on vascular attenuation. Radiology 2005; 236:661–665
47. Utsunomiya D, Awai K, Sakamoto T, et al. Cardiac
16-MDCT for anatomic and functional analysis: assessment of a biphasic contrast injection protocol.
AJR 2006; 187:638–644
48. Wilting JE, Timmer J. Artefacts in spiral-CT images
and their relation to pitch and subject morphology.
Eur Radiol 1999; 9:316–322
49. Maintz D, Seifarth H, Raupach R, et al. 64-slice
multidetector coronary CT angiography: in vitro
evaluation of 68 different stents. Eur Radiol 2006;
16:818–826
50. Schepis T, Koepfli P, Leschka S, et al. Coronary artery stent geometry and in-stent contrast attenuation
with 64-slice computed tomography. Eur Radiol
2007; 17:1464–1473
F O R YO U R I N F O R M AT I O N
This article is available for CME credit. See www.arrs.org for more information.
AJR:189, September 2007
591