AJR 2005; 184:1402-1412
© American Roentgen Ray Society
MDCT Evaluation of the Coronary Arteries, 2004: How We Do ItData Acquisition, Postprocessing, Display, and Interpretation
Leo P. Lawler1,
Harpreet K. Pannu and
Elliot K. Fishman
1 All authors: Department of Radiology and Radiological Science, Johns Hopkins
University, 601 N Caroline St., Rm. 3254, Baltimore, MD 21287-0801.
Received February 12, 2004;
accepted after revision September 15, 2004.
Address correspondence to L. P. Lawler.
L. P. Lawler and H. K. Pannu contributed equally to this article.
Abstract
OBJECTIVE. Cardiac CT is rapidly becoming part of clinical practice.
The objective of this article is to discuss and illustrate the current
practice of coronary artery MDCT, including data acquisition, postprocessing,
image display, and interpretation. The practice described reflects our
experince with a series of patients referred in routine clinical practice.
CONCLUSION. The reader should gain an insight into the current
clinical application of coronary artery CT.
Introduction
With the improving temporal and spatial resolution of mechanical MDCT
scanners, coronary artery imaging is entering the clinical realm
[15].
Although 16-MDCT scanning diminishes motion artifact and the limits of
single-detector technology, it remains an emerging, imperfect technology that
requires careful attention to technique for optimal performance and clinically
useful imaging. Future larger detector arrays, changing tube technology, and
simplified postprocessing platforms will enhance and change coronary CT
angiography protocols. There is not yet any universal agreement on the method
of performance or interpretation of such studies or their role in the clinical
algorithm, although the demand for coronary CT angiography continues
unabated.
In this article, we share our state-of-the-art approach to CT angiography
with the goal of providing a framework for those initiating coronary MDCT
angiography in clinical practice. We cover data acquisition, processing, and
postprocessing and suggest an approach to image review, interpretation, and
reporting. Although the emphasis is on native vessel imaging, the principles
also apply to bypass graft study. Please note that the article is based on the
evidence currently available and although the practice reflects a single
approach at our institution, it is of course not the only way to perform
coronary CT angiography.
Patient Selection
In part because of the lack of large outcome study results, there is not
yet a consensus statement on the role of CT angiography in the current cardiac
evaluation algorithm. There are no universally accepted absolute indications
for cardiac CT angiography, and the patient population and principle of
practice at present appropriately reflect those with a low pretest probability
of coronary artery disease for which the high negative predictive value of the
test can be applied.
Patients are referred for coronary CT angiography if their doctor believes
the level of suspicion does not justify an invasive test but there is concern
for unsuspected disease that may change management. This group usually
reflects asymptomatic patients with a low pretest probability of coronary
artery disease and often no cardiac history. They do have identifiable
Framingham risk factors, the most common of which are family history and
raised cholesterol. Select patients will be referred for coronary CT
angiography when they are deemed high risk for conventional invasive
angiography or when invasive imaging has failed. Younger patients are also
referred if there is suspicion of aberrant origin of the coronary arteries and
if findings from catheter angiography or echocardiography are
nondiagnostic.
We do not perform coronary CT angiography in those suspected of having
acute coronary syndromes, for which conventional angiography remains the
standard and intervention is anticipated. Significant disease can be excluded
when normal vessels are seen, although confidence in predicting unstable
plaque and coronary events falls with increasing atherosclerotic disease
burden. We do not perform coronary CT angiography in those with a heavy burden
of calcified plaque because estimating the degree of stenosis at the site of
calcified plaque will be difficult due to blooming artifact and because a
catheter angiogram will still be necessary. We do not proceed with studies in
patients with a heart rate of more than 70 beats per minute (bpm) or with
significant arrhythmia because the resulting images are poor. Breath-hold
difficulties and the inability to remain supine and motionless are relative
contraindications. We do not do plaque characterization studies in routine
practice, although it is applied in a research capacity.
At present there is no specific reimbursable code for dedicated coronary CT
angiography, but there are number of working groups engaged in this issue as
the use becomes more widespread. Some patients pay out of pocket for the
study, although the examination can be coded under a contrast study of the
chest with a charge for 3D reconstruction.
Patient Preparation
The main preparation involves slowing the patient's heart rate to
approximately 60 bpm because this lower heart rate increases the relative
proportion of the cardiac cycle spent in diastole and limits motion artifact
[6,
7]. During the patient
assessment, one should note that the heart rate frequently rises with the
anxiety of being placed on the scanner and with the injection of contrast
material. A decrease in heart rate has also been noted during the initial
seconds of breath-holding. At our institution, generally oral ß-blockers
are prescribed by the referring doctor 23 days before scanning and
organized at the time of scheduling. Some sites also give ß-blockers at
the time of the study as well.
Sample ß-blocker protocols include giving 50 mg of metoprolol orally 1
hr before scanning or 5 mg of metoprolol IV a few minutes before the study.
Calcium channel blockers are used if ß-blockers are contraindicated.
Contraindications for ß-blocker therapy include asthma, atrioventricular
conduction block, heart failure, diabetes, and Raynaud syndrome. Three ECG
leads are placed over the patient as specified by the manufacturer of the
scanner to obtain an ECG tracing on the scanner console.
Contrast Injection and Scan Delay
The patient's history is reviewed to exclude the typical contraindications
for contrast administration such as renal insufficiency and allergic reaction.
Informed consent is obtained at our institution. The cardiovascular effects of
contrast media are in part related to their osmolality and include
abnormalities of conduction and contractility, although changes in hemodynamic
parameters are generally not significantly different among the various
nonionic contrast media [8,
9].
We use a dual-head power injector (Stellant, Medrad) to inject 100 mL of
isosmolar nonionic contrast material through an 18- to 20-gauge needle into an
antecubital vein at a rate of 3.54 mL/sec; the injection is followed
immediately by a 25-mL saline flush. The purpose of a saline flush is to
diminish beam-hardening contrast artifact within the right ventricle that
obscures the right coronary artery. It also facilitates delivery of the entire
contrast volume in a short bolus
[10].
At our institution, the scan delay is empirically timed to coincide with
the end of the contrast injection and the beginning of the saline flush. This
delay is approximately 25 sec. An alternative method of determining the scan
delay is to use bolus-tracking or test bolus techniques. With the former
technique, a region of interest is drawn at the ascending aorta and scan
acquisition begins when a predefined threshold Hounsfield unit of contrast is
reached after power injection. With the test bolus technique, a small contrast
bolus is administered while scanning a fixed level in the ascending aorta.
From this, a curve of the contrast density rise and fall is generated.
Assuming this test bolus will reflect the behavior of the main contrast
volume, the scan timing may be deduced from the peak density value on the
curve. A good technique shows the highest contrast in the left ventricle and
coronary arteries with less density in the right ventricle and pulmonary
arteries.
Scanning Protocol
The scan coverage is from the carina through the heart base in a
cranial-to-caudal direction. The aortic arch is the superior extent if bypass
grafts or internal mammary arteries are also assessed. Some advocate an
initial unenhanced study of the heart either by routine scanning or as per a
coronary calcification scoring protocol. This study is performed to identify
and exclude those patients whose burden of calcified plaque will significantly
compromise any coronary CT angiography.
The thinnest detector collimation possible is selected, typically in the
range of 0.751 mm on a 16-MDCT scanner (Sensation 16, Siemens Medical
Solutions). The fastest gantry rotation time possible is selected, a typical
value being 420 msec. Typical table feed is 2.8 mm per gantry rotation, which
gives a lower pitch than routine scanning. The peak kilovoltage is 120 kVp,
but it can be lowered to 100 kVp for thin patients. The effective
milliampere-second setting selected is higher (
500 mAs) than for routine
non-ECG-gated CT studies to decrease image noise from partial scan
reconstruction and small slice width. ECG pulsing can be used whereby the tube
output is decreased during systole and increased during diastole when target
images will be reconstructed, reducing radiation dose by approximately 50%.
The effective dose will range from 7 to 10 mSv and is higher in women
[11]. The dose is higher than
routine thoracic CT angiography and higher than conventional angiography.
Retrospective ECG gating is used for helical coronary CT angiography
studies performed on an MDCT scanner. With this method, the scanning data and
ECG tracing are recorded simultaneously but independently. It allows the
scanning data to be acquired throughout the cardiac cycle for subsequent
reconstruction during specified periods of the cycle. The entire heart is
imaged as a volume for subsequent 3D manipulation. With ECG gating, there is
motion artifact if the heart rate is fast or the rhythm is irregular. In
select cases of patients with high heart rates, a non-ECG-gated routine CT of
the heart with thin collimation and fast gantry rotation speed may be
sufficient to answer the clinical question (e.g., aberrant vessel origin or
high-grade proximal stenosis). However, ECG-gated studies are the optimal
method for clinical practice in most patients.
Scan Reconstruction
Slice Width and Field of View
The scan is reconstructed with a slice thickness of 1 mm with 2550%
overlap as is the convention for optimal 3D imaging. A medium-smooth
reconstruction algorithm is used, similar to that for most routine chest and
abdominal studies. If a coronary artery stent is present, a second
reconstruction with a slightly sharper algorithm (higher spatial frequency)
can be done through the area of the stent. The field of view is coned down to
the heart to improve spatial resolution.
Temporal Resolution and Reconstruction Window
Images are reconstructed for each position in the z-axis during
the time of least cardiac motion or diastole. The scan is reconstructed using
only a small portion of the cardiac cycle, the duration of which represents
the time needed to acquire the necessary datathat is, the temporal
resolution. The temporal resolution of currently available MDCT scanners for
coronary artery imaging is approximately 105210 msec. The ECG tracing
is reviewed, and the reconstruction window is defined by the operator using
either a relative (percentage) or fixed (absolute) time delay from the R wave.
The relative delay divides the R-R interval into percentage increments from 0%
to 100% (Fig. 1). Image
reconstruction is started with a certain delay from the prior R wave. The
delay is defined as a percentage of the R-R intervalfor example, 60%.
These percentage numbers specify the point in the cardiac cycle at which image
reconstruction is started. The length of time used to reconstruct the data set
depends on the temporal resolution of the scanner. If multiple reconstructions
are done with varying delays, there is overlap in the image sets when the
reconstruction increment is less than the temporal resolution of the
scanner.

View larger version (8K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1. ECG trace shows relative delay method for scan
reconstruction. R-R interval is divided into percentage increments from 0% to
100%: 0% at first R wave and 100% at second R wave. Image reconstruction is
started with a certain delay from the prior R wave. The delay is defined as a
percentage of the R-R intervalfor example, 60%.
|
|
A fixed or absolute time from the R wave can also be used to reconstruct
images (Fig. 2). With the
absolute delay method, a fixed time delay after the R wave is used to start
reconstructionfor example, 400 msec from the R wave. With the absolute
reverse method, a fixed time before the next R wave is used to start
reconstructionfor example, 400 msec before the R wave. The relative
delay or absolute reverse method is usually used for reconstruction. We
typically use the relative delay method.
One may reconstruct at different percentage increments a single slice
through the mid right coronary artery to find the image of least motion. This
is the most diastolic phase of the cardiac cycle and may be partly deduced
from the state of wall thickness, chamber volume, and the state of
atrioventricular valve closure. Once the image with the least artifact is
found, that percentage delay may be used to reconstruct the entire cardiac
volume. If there is motion artifact involving one particular vessel,
additional reconstructions with different delays are performed. Kopp et al.
[5,
12] found that the right
coronary artery was best seen early in diastole at 40% of the R-R interval,
the left circumflex artery was best seen in mid cycle, and the left anterior
descending artery was best seen at 6070% of the R-R interval. For
coronary CT angiography studies, reconstructions at 10% increments from
4080% will likely provide optimal images to evaluate all the
vessels.
This process of creating multiple reconstructions is usually done on the
scanner itself and therefore can delay scanning of other patients. It can be
done after hours or, to help with workflow, the technologist can be instructed
to perform multiple reconstructions for all patients so that all are available
for the radiologist to review. Newer scanners will allow processing to be done
on a separate workstation so that scanning may be uninterrupted. In our
practice, standardized incremental 10% reconstructions are done by the
technologist for all studies throughout the cardiac cycle. For example,
reconstructions are done at 10% increments from 090% of the cardiac
cycle. This helps with workflow but does create large amounts of data that
require a separate workstation for storage until analysis is completed. The
advantage of this approach is that one also obtains systolic and diastolic
sampling of the cardiac cycle for functional imaging, such as calculation of
ventricular ejection fraction and assessment of wall motion.
Some scanners allow a series of interval reconstructions at a single level
to be reviewed at once so the cardiac phase of interest may be reconstructed.
Newer tools with cine 3D (sometimes called 4D) of all cardiac phases also
allow one to pick the most diastolic or systolic images of interest. It is
helpful if the raw data of the study are not deleted until the study has been
interpreted and no additional information is needed. Having access to the raw
data is helpful for cases in which poor image quality is secondary to poor ECG
gatingfor example, when there is a poor ECG trace or arrhythmia. Review
of the raw data is done at the workstation before the raw data are erased. ECG
editing allows you to review where the data reconstruction was relative to the
cycle of the ECG trace. If you think the portion of the cycle used is
incorrect, you can manually change it. Any additional beats that are used for
image reconstruction may be deleted.
Single- or Multisegment Data Reconstruction
Temporal resolution is determined largely by gantry rotation time but may
be enhanced by segmental reconstruction
[13]. Single reconstruction is
the preferred method and is successful in patients with low heart rates. A set
of images, equal to the total detector width, is reconstructed from data
acquired during one gantry rotation and one cardiac cycle
(Fig. 3). The image is
reconstructed using data acquired during approximately 240° of the gantry
rotation (180° + a portion of the detector fan angle) rather than the
entire 360°. The temporal resolution is therefore improved to half the
gantry rotation time [14]. If
the rotation time is 420 msec, the temporal resolution is approximately 210
msec, although it may vary between 210 and 280 msec.
In patients with low heart rates (
50 bpm), single-sector
reconstruction in mid-diastole is preferred. In those with a high heart rate
(
70 bpm), image reconstruction cannot be completed in the short diastolic
time available without giving artifacts
[13]. Therefore, multisegment
reconstruction is done using small segments of data acquired during two or
more cardiac cycles [13,
15]
(Fig. 4). Multiple segments of
data are combined to make the final image at end-systole. The resulting
temporal resolution equals the gantry rotation time divided by 2N,
where N is the number of cycles. If the rotation time is 420 msec,
the temporal resolution is 105 msec. However, the z-axis spatial
resolution can decrease if there are gaps in the acquired data
[13]. The maximum number of
cycles typically used is two or three, beyond which image quality falls. To
avoid loss of z-axis resolution, we prefer single-segment
reconstruction over multisegment reconstruction and have found that prescan
optimization of the patient's heart rate is more helpful than postscan
processing to get a good quality scan.
Problem Solving
Occasionally, the study is not optimal because of cardiac motion artifacts,
contrast artifact, and poor contrast bolus. The optimal study is obtained when
the patient has a regular sinus rhythm.
In cases in which the cardiac rhythm is irregular and there is artifact due
to cardiac motion, postprocessing ECG editing can be attempted to improve the
quality of the study. Such editing is limited to one or two beats because
greater manipulation will lead to gaps in the data. The technique of editing
varies among scanner vendors and can be obtained from the applications
specialist of your scanner's vendor. Cardiac motion is identified when there
is stairstep artifact in the cardiac or vessel contour on reconstructed images
(Fig. 5A,
5B). This stepoff between two
sets of reconstructed images occurs when the heart is not in the same position
from the time the first set is acquired to the time the second set is
acquired. The ECG trace is reviewed to ensure that the bars specifying the
part of the cardiac cycle during which the images are being reconstructed are
positioned in diastole. An aberrant R wave that occurs sooner than
expectedshortening the anticipated R-R intervalcan result in the
reconstruction bar being positioned in systole (Figs.
5A,
5B and
6A,
6B,
6C,
6D). The image slices with
artifact are identified, and the corresponding reconstruction bar on the ECG
trace is deleted.

View larger version (5K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5B. Images of 52-year-old man show stairstep artifact. ECG trace
shows early R wave(arrowhead). Relative delay method has been used to
reconstruct images. Reconstruction is in diastole for first cardiac cycle.
Reconstruction overlaps systole for next cycle because of early R wave. Arrow
shows start of reconstruction of second image set. ECG trace can be edited to
delete this reconstruction. Double-headed arrows show relative delay.
|
|

View larger version (117K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6B. Images of 60-year-old woman show stairstep artifact and
effect of editing ECG trace. ECG trace from scan shown in A reveals
second reconstruction set for images 1324 (arrow) overlapping
systole. Second and third reconstruction sets are in different points of
cardiac cycle. Third reconstruction set is for images 2531.
|
|

View larger version (38K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6C. Images of 60-year-old woman show stairstep artifact and
effect of editing ECG trace. ECG trace shown in B was edited to delete
second reconstruction (arrow). Images 1331
(arrowhead) are now reconstructed from next cardiac cycle.
|
|

View larger version (47K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6D. Images of 60-year-old woman show stairstep artifact and
effect of editing ECG trace. Coronal multiplanar reconstruction CT image shows
stairstep artifact (arrow) is gone as a result of editing of ECG
trace (C).
|
|

View larger version (25K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8B. Images of coronary artery in 56-year-old man obtained using
postprocessing techniques. Multiplanar reconstruction CT image shows
centerline vessel straightening of right coronary artery
(arrowhead).
|
|
If there is increased beam-hardening artifact from contrast material in the
right heart that affects the right coronary artery in particular, a saline
bolus after contrast injection or slightly increasing the scan delay may be
helpful. If the contrast bolus is poor, the injection speed can be increased.
Artifact may occur because of respiratory motion or image noise. Limiting the
scanning time to the minimum necessary can help reduce the likelihood of
respiratory motion. For scans with increased image noise, a smoother
reconstruction algorithm and slightly thicker slice reconstruction of 1.5 mm
can decrease the amount of noise. However, the spatial resolution also
decreases.
Data for Postprocessing
A large MDCT data set is obtained from sampling the entire cardiac cycle,
from which selective postprocessed image sets must be generated for greatest
diagnostic yield. Although an infinite number of data sets of the cardiac
cycle may be generated from the raw data, the needs of coronary CT angiography
are limited to the diastolic frames.
We use a workstation (Leonardo, Siemens Medical Systems) that permits
real-time axial, multiplanar reconstruction, maximum-intensity-projection, and
volume-rendering techniques
[16]. However, these tools are
available on the workstations of most scanner vendors and are present on many
third-party stand-alone workstations. To this workstation, we typically send a
range of reconstructions10%, 20%,... 90%representing the
systolic and diastolic sampling of the entire R-R interval of the cardiac
cycle. At present, we do this because there is independent motion of the right
and left coronary arteries. The right coronary artery may show greater motion
in late diastole, whereas the left coronary artery branches may show greater
motion in early diastole. Therefore, more than one reconstruction may be
necessary to optimally show both vessels.
There is also interpatient variation, so that the reconstruction window in
the cardiac cycle that is optimal for the right coronary artery in one patient
may not be suitable for that vessel in another patient. Therefore, the point
in the cardiac cycle chosen for image reconstruction has to be customized for
each individual patient and for each coronary artery
[5,
12]. Although the 4070%
reconstructions suffice in most patients, the full range of data also allows
for supplemental volume measurement, functional imaging, and 4D review in
select cases where attendant pathology makes this of clinical value. Future
software upgrades promise to make data reconstruction and selection of optimal
postprocessing phase simpler and even semiautomated.
The workflow varies among institutions. At the outset, it is good for the
radiologist to be involved in all aspects of data acquisition and
postprocessing to learn and to help technologists understand and develop the
imaging protocol. In most established practices, the technologist does most of
the initial data processing that requires reconstruction of image sets
throughout the cardiac cycle. Newer technologies may automate much of this
handling of the raw data. Postprocessing is done by those available and
willing to do it.
At our institution, the radiologists perform all the 3D postprocessing, in
part because of the complexity of the conditions and clinical questions but
also to allow direct real-time consultation with referring doctors. However,
perhaps in most institutions, it is the technologists who are trained in the
use of the workstations, although the feasibility of this approach requires a
clear consensus among radiologists about the types of multidimensional images
and perspectives required in individual patients.
Viewing Perspectives
Unlike conventional angiography, there are countless multidimensional
viewing angles available with MDCT angiography. For efficiency, we limit these
to the standard and patient-specific views. The standard views are those
favored by most clinicians and similar to conventional angiography: the right
anterior oblique view of the right coronary artery, the left anterior oblique
view of the left coronary artery, and the left lateral oblique view of the
left anterior descending and circumflex arteries (Fig.
7A,
7B,
7C).

View larger version (158K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7A. Coronary artery anatomy in 56-year-old man on 3D MDCT images
obtained with volume-rendering technique. Left lateral view shows left
anterior descending (arrow) and circumflex (arrowhead)
arteries.
|
|

View larger version (119K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7B. Coronary artery anatomy in 56-year-old man on 3D MDCT images
obtained with volume-rendering technique. Superior view shows origins of right
(arrow) and left (arrowhead) main coronary arteries.
|
|
A superior perspective on MDCT angiography or a perspective in the plane of
the coronary origins provides a clear depiction of the site and patency of the
right and left coronary origins and provides a satisfactory view of the left
main coronary artery. Two angled right anterior oblique perspectives usually
depict the proximal and middle portions of the right coronary artery. A left
lateral view depicts the left anterior descending artery, but a superior
perspective is required for diagonal and septal perforating branches. The
circumflex artery is best seen using a left lateral or posterior oblique
projection for the proximal and mid portions. The posterior descending artery
in the atrioventricularinterventricular groove from the right coronary
artery or circumflex artery is best appreciated with an inferior
perspective.
The patient-specific views are those dedicated perspectives that best
depict individual pathology to best effect. Because stenotic changes in
caliber are frequently noncircular and irregular, one will find that at least
two perspectives of each artery segment are required to confidently exclude
caliber change.
Postprocessing Techniques
Current postprocessing uses a combination of 2D and 3D techniques
[4,
1719].
Two-dimensional tools include routine axial planar imaging and multiplanar
reconstruction. Multiplanar reconstruction requires little computer power and
reorders the voxels into an arbitrary straight or curved imaging plane
(Fig. 8A). Curved multiplanar
reconstruction images may be generated through user definition of the
reconstruction plane from a series of planar source images. This technique is
subject to user error defining the vessel center, however, and only one branch
at a time may be depicted. Centerline tools have sought to minimize this error
and facilitate semiautomated drawing of curved multiplanar reconstructions.
With this tool, the user plants a seed point in the region of the vessel and
it automatically defines a plane in the center of the lumen contrast. This
approach is limited in its application for small-caliber vessels and has
difficulty discriminating contrast material and calcification to define the
vessel center. Curved vessel images obtained with centerline tools may be
straightened to show the relative caliber at different anatomic points
(Fig. 8C).
Three-dimensional tools include maximum-intensity-projection and
volume-rendering techniques [4,
1720].
Maximum intensity projection is a projection technique akin to conventional
angiography that preferentially displays density values from a visual ray
above a chosen threshold value. Slab maximum intensity projection with editing
planes in front and behind a vessel are an efficient means by which to produce
high-contrast images over long vessel segments
(Fig. 8C). Volume rendering is
a computer-intensive technique with high fidelity to the acquired data that
displays the range of density values within a voxel and confers opacity and
depth to the image. We try not to vary segmentation thresholds or transfer
functions because this can lead to diminished intra- and interscan consistency
by obscuring or creating stenoses
[19,
21,
22]. We do not use
shaded-surface display or virtual angioscopy
[15], although the latter may
have some value in evaluating patency in the setting of heavily calcified
vessels or stents.
Gray-scale images suffice, although color images have an esthetic appeal
and can better separate vessels from underlying cardiac chamber contrast
material. The center and width are adjusted to limit the blooming artifact of
large calcifications or coronary stents.
Vessel Review Sequence
It is important to note that though a small field of view around the heart
may be generated for the CT angiography, the raw data contain the entire
thorax, which must be reviewed in the axial plane with standard window
settings to evaluate for noncardiac abnormalities or disease. Second, we
assess for noncoronary-artery cardiac findings of
significancefor example, valvular or pericardiac abnormalities.
Myocardial thinning, contour change, or attenuation change may reflect prior
infarct and is noted. We also comment on gross abnormalities of chamber size
and shape.
An initial brief 2D axial planar and volume-rendered 3D review of the
coronary arteries will greatly facilitate subsequent dedicated vessel
interpretation by giving the radiologist an overview of vessel course so that
segmentation editing may be more correctly applied. American Heart Association
nomenclature is applied [2,
5,
23], which acknowledges three
main coronary arteries: left anterior descending, circumflex, and right
coronary with right, left, or codominant systems
(Fig. 9). The coronary anatomy
is divided into 29 segments, although in practice only 1215 are
routinely used, and disease is classified as one-, two-, three-vessel or left
main distribution. We have established a style and language of reporting in
consultation with our referring clinicians.
Dedicated assessment of the four main epicardial vessels is performed in a
sequential manner: right coronary artery, left main, left anterior descending,
and circumflex arteries. Side-branch vessels are not usually evaluated unless
they are of a caliber greater than 2 mm, but they are used for localization.
Initially, vessel origin, course, and relative dominance are noted.
Distribution and burden of calcification or plaque are evaluated, and main and
branch vessel opacification is documented. At this point, many vessels may be
defined as assessable or nonassessable. The percentage of assessable vessels
will vary between 68% and 94% depending on patient and acquisition factors
[24].
Each vessel is initially viewed using volume rendering. Volume-rendering
images give an overview of the entire vessel and are useful to depict and
correctly name the relative sites of calcification, stenoses, and stents (Fig.
10A,
10B). By seeing the vessel
course initially in 3D, one may more efficiently and accurately assess caliber
using 2D techniques. The simplest means to view long vessel segments or
vessels in their entirety is to place a second parallel plane (slab editing)
posterior to the vessel of interest and apply slab multiplanar reconstruction
or maximum intensity projection
[2,
19,
21] (Fig.
6A,
6B,
6C,
6D).

View larger version (83K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 10A. 54-year-old man with coronary artery disease. Volume-rendered
left lateral view shows calcified left anterior descending artery with short
segment that has proximal stenosis (arrow) of greater than 50%.
|
|

View larger version (89K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 10B. 54-year-old man with coronary artery disease.
Maximum-intensity-projection image shows left lateral view of calcified left
anterior descending artery with short segment that has proximal stenosis
(arrow-head) of greater than 50%.
|
|
It remains controversial which is the most accurate postprocessing
technique for caliber quantification
[4,
1720].
We find at present stenoses are frequently detected on the volume-rendering
images but are best confirmed and characterized using some form of planar
technique (Figs. 10A,
10B and
11A,
11B). Such techniques are
closest to the original data acquired and have limited reconstruction
artifact. Although absolute caliber measurements can be obtained from 2D and
3D data sets, we do not routinely report them at this time because none of the
tools has proven accuracy as yet. Thus, as in most conventional angiography
reports, we prefer to provide an impression of percentage of narrowing (in
quartiles) relative to the maximum diameter seen.

View larger version (106K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 11A. 72-year-old woman with coronary artery disease. Coronal
volume-rendered view (A) and coronal maximum-intensity-projection view
(B) show stenosis (arrowhead) of greater than 70% that is due
to focal noncalcified plaque in proximal right coronary artery
(arrow).
|
|

View larger version (121K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 11B. 72-year-old woman with coronary artery disease. Coronal
volume-rendered view (A) and coronal maximum-intensity-projection view
(B) show stenosis (arrowhead) of greater than 70% that is due
to focal noncalcified plaque in proximal right coronary artery
(arrow).
|
|
Stenoses less than 50% are not considered hemodynamically significant, but
they are not necessarily clinically benign: Plaques that underlie them may
progress or rupture acutely. Significant stenoses are considered those greater
than 50% [2,
5]. High-grade stenoses are
generally accepted to be those with more than 7075% narrowing
[23,
25]. We do not attempt any
quantification of severely calcified vessels or stented vessels where the
lumen cannot be confidently discriminated, and 1- to 1.5-mm vessels cannot be
reliably assessed. Although an association of plaque characteristics and
sonographic features has been shown
[2,
15], there are no data on the
ability of clinical CT to detect or quantify noncalcified plaque; thus, plaque
characterization is limited to site and calcified or noncalcified descriptors
using multiplanar reconstructions. Separate dedicated calcium scoring
examinations are performed when specifically requested.
The typical final report describes the pattern of dominance and the burden
of atherosclerotic change. Lesions are characterized by location, length, and
severity, and associated plaque or calcification is described. Internal
mammary arteries, which are invariably patent, are documented. When
specifically requested, calcification scores and functional information on
wall thickness, motion, and ejection fraction are provided. Any limitations of
vessel visualization are acknowledged. Incidental noncardiac and noncoronary
cardiac findings are discussed. At present, reports are narrative; however,
using a new voice-recognition dictation system, we are formulating a standard
reproducible format to cover the technique and findings. We send photographic
prints to the referring clinician and digital static JPG or TIFF images and
dynamic audio-video inter-leave files when requested.
What to Expect
The left main, proximal left anterior descending, circumflex, and right
coronary arteries are easiest to assess, and good quality images are
reproducible. Approximately 11 cm of the left anterior descending and the
right coronary arteries and 8 cm of the circumflex artery may be displayed and
over 70% may be motion-free
[24].
Successful evaluation is caliber-dependent, and, in particular, interpretation
of 3- to 4-mm proximal vessels is superior to distal and branch 1- to 2-mm
vessels [3,
4]. The more mobile distal
right coronary artery and circumflex artery within the atrioventricular groove
can prove more difficult, and 2532% of arteries will be excluded from
analysis because of motion artifact or calcification and poor opacification
[2,
5,
25]. An overlapping coronary
vein may also limit interpretation for certain portions of the circumflex
artery.
Our experiencethe ability to detect significant stenoses in
7095% with specificities ranging from 80% to 95%reflects that of
the literature, and we find a good negative predictive value with high
confidence in excluding disease when normal vasculature is depicted.
Determining the patency of stents and of heavily calcified coronary arteries
remains problematic in small-caliber vessels, although it may be suggested
from more distal opacification
[24,
26] (Fig.
11A,
11B), and eccentric
calcification can be documented on orthogonal vessel imaging
(Fig. 12). MDCT angiography
performs well in evaluating the less mobile venous and arterial bypass grafts,
but these patients frequently have advanced native vessel disease beyond the
anastomoses that cannot be fully assessed (Fig.
13A,
13B).

View larger version (112K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 12. 63-year-old man with coronary artery disease. Right sagittal
maximum-intensity-projection image of left anterior descending artery shows
focal inferior eccentric calcification (arrow).
|
|

View larger version (99K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 13A. 60-year-old man undergoing assessment after coronary artery
bypass grafting. Volume-rendered anterior view shows saphenous vein bypass
grafts (arrowheads) from aneurysmal aorta (A) and left
internal mammary bypass graft (arrow).
|
|

View larger version (120K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 13B. 60-year-old man undergoing assessment after coronary artery
bypass grafting. Volume-rendered left lateral view shows saphenous vein
(arrowhead) and left internal mammary (arrow) bypass
grafts.
|
|
We have found CT angiography successful and superior to catheter studies in
evaluating aberrant vessel, origin course, and caliber, and our CT angiography
findings have shown high concordance with surgical findings (Fig.
14A,
14B). We perform coronary CT
angiography in the context of a broad cardiovascular evaluation, and a
discordant negative result with a high pretest probability of disease still
requires additional cardiovascular evaluation.

View larger version (179K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 14A. 30-year-old woman referred for evaluation of aberrant
coronary artery origin. Volume-rendered superior oblique image shows common
origin of aberrant right (short arrow) and left main
(arrowhead) coronary arteries from left aortic sinus. Note left
anterior descending coronary artery (long arrow).
|
|

View larger version (118K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 14B. 30-year-old woman referred for evaluation of aberrant
coronary artery origin. Maximum-intensity-projection superior oblique view
shows common origin of aberrant right (arrowhead) and left main
(arrow) coronary arteries from left aortic sinus.
|
|
Conclusion