Potential indications for coronary CTA are many and continue to evolve. Recently, the American College of Cardiology Foundation Quality Strategic Directions Committee Appropriateness Criteria Working Group, the American College of Radiology, and others have published a joint review on the appropriateness of cardiac CTA.¹ This report scored the appropriateness of cardiac CTA in several clinical applications on a scale from 1 to 9 based on the available literature. The information and experience available with cardiac CTA at this time are not robust enough to support categorical recommendations and guidelines. This is so because the technology continues to change rapidly, there are not enough prospective studies on different groups of patients, and there are no outcome studies on the prognostic value of cardiac CTA. Therefore, the appropriateness criteria are not guidelines but rather temporary recommendations until official guidelines are published.

Among the potential indications for cardiac CTA, there are four clinical groups. The first group includes patients with an indication for invasive catheterization when the procedure is anticipated to be technically demanding (e.g., ascending aortic aneurysm, atherosclerotic arch, anticoagulant therapy), patients with a prior incomplete catheterization, and patients with suspected coronary anomalies. For patients in this group, cardiac CTA may be the optimal imaging modality and in certain cases the only one.

The second group consists of patients with an intermediate pretest probability for obstructive coronary artery disease. This includes patients with atypical symptoms, acute chest pain of uncertain origin, and inconclusive stress test or thallium scan as well as candidates for valve or noncardiac surgery. The purpose of cardiac CTA in this group is to exclude significant coronary stenosis, and a normal scan may obviate the need for a catheter-based intervention.

The third group includes patients with known coronary artery disease (e.g., patients after stent placement or bypass grafting) who present with a specific question, such as stent or graft patency. This group of patients tends to have frequent complaints; traditional noninvasive tests have limited predictive accuracy, and invasive procedures are commonly employed. Here, exclusion of significant stenosis may prevent unnecessary catheter-based intervention.

The fourth group is controversial and consists of asymptomatic individuals at risk for coronary artery disease. On the one hand, CT is able to identify subclinical atherosclerosis, and the findings themselves may lead to improved compliance with preventive regimens. On the other hand, CTA involves administration of contrast material and radiation exposure, and we currently have not established an approach to dealing with such findings.

Contraindications

There are only a few contraindications to cardiac CTA, the main being a contraindication to administration of iodinated contrast agents. Injection of contrast material is mandatory for visualization of the coronary arteries.

Inability to follow breath-hold instructions (8 to 20 seconds, depending on the scanner used) may also be considered a contraindication. Breathing-induced motion artifacts may render a study unassessable.

Irregular heart rate is considered a relative contraindication, depending on scanner type and available algorithms for dealing with variable R–R intervals.

Excessive calcium within the wall of the coronary arteries is sometimes considered a contraindication to cardiac CTA. It may lead to overestimation of stenosis severity and prompt unnecessary intervention.

Technique Description

Cardiac CTA is a contrast-enhanced spiral study performed with ECG gating to eliminate cardiac motion. It is technically complex and requires high spatial resolution, high temporal resolution, good low-contrast resolution, intravascular contrast enhancement, and short scanning time. Advanced scanners have increased coverage, faster gantry rotation time, and sophisticated algorithms for reconstruction of partial scan data or combination of data from various phases (multicycle reconstruction) to further reduce temporal resolution.

Compared with invasive catheterization, cardiac CT evaluates the coronary arteries noninvasively, displays its wall as well as the lumen, and allows quantification and characterization of atherosclerotic plaques even before there is luminal narrowing.

Initially, scout images are acquired to determine anatomic span. For native coronary arteries, the CTA scan extends from the level of the carina to the base of the heart. In patients who have had bypass surgery, scans are started at the level of the clavicles. The scan is performed during a single breath-hold with timed injection of contrast material to achieve optimal enhancement of the coronary arteries.

Images are typically reconstructed from data acquired at 70% to 75% and 40% to 45% of the R–R interval. If needed, additional phases are reconstructed to find a “quiet” (motionless) phase.

Pitfalls and Solutions

In CT, the term artifact applies to any systematic discrepancy between the CT numbers in the reconstructed image and the true attenuation coefficients of the object. CT images are inherently prone to artifacts because the image is reconstructed from multiple independent detector measurements. The reconstruction algorithm assumes that all these measurements are consistent, so any error of measurement will usually be reflected as an error in the reconstructed image.²^,³

Four types of artifact can occur: streaking, due to an inconsistency in a single measurement; shading, due to a group of channels or views deviating gradually from the true measurement; rings, which are due to errors in an individual detector calibration; and distortion, due to the helical reconstruction.

Artifacts can seriously degrade the quality of CT images, sometimes to the point of making them diagnostically unusable. Furthermore, accuracy of cardiac CTA for detection of stenoses depends highly on image artifacts,³ which are a major cause of false-positive and false-negative interpretations.

To optimize image quality, it is necessary to understand why artifacts occur and how they can be prevented (Table 35-1). CT artifacts originate from a range of sources. Physics-based artifacts result from the physical processes involved in the acquisition of CT data. Patient-based artifacts are caused by such factors as movement of the patient or the presence of metallic materials inside or on the patient.

TABLE 35-1 Artifacts, Their Appearance and Causes, and Measures to Avoid Them

Careful preparation of the patient and optimum selection of scanning parameters are, therefore, important factors in avoiding CT artifacts.

Patient Motion (Blurring)

Motion of the patient can cause misregistration artifacts, which usually appear as shading or streaking in the reconstructed image (Fig. 35-6). Steps can be taken to prevent voluntary motion (due to movement or breathing during the scan), but some involuntary motion may be unavoidable. Prescan instructions about the expected breath-hold are critical to minimize motion artifacts as well as to decrease scan time. Involuntary motion artifacts may be caused by heart rate irregularities during the scan and will appear as blurring or stair-step artifacts. Looking for a motion-free phase can sometimes help improve visualization of the coronary arteries (Fig. 35-7).

FIGURE 35-6 Two pairs of axial slices (in lung and soft tissue windows) show motion artifacts due to breathing (A and C), causing blurring and streaking of the coronary arteries (B and D).

FIGURE 35-7 Involuntary motion artifacts may be caused by heart rate irregularities during the scan. The right coronary artery and, to a lesser extent, the circumflex artery are more susceptible to motion artifacts. Comparison of axial slices (A) and oblique coronal maximum intensity projection images (B) from phases 75% (top) and 0% (bottom) shows improved right coronary artery visualization with phase 75% (arrows).

Stair-Step Artifacts

Stair-step artifacts appear as horizontal lines through the image, visible especially around the edges of structures in multiplanar and three-dimensional reformatted images, when wide collimations and non-overlapping reconstruction intervals are used. They are less severe with helical scanning, which permits reconstruction of overlapping sections without the extra dose to the patient that would occur if overlapping axial scans were obtained. Stair-step artifacts are virtually eliminated in multiplanar and three-dimensional reformatted images from thin-section data obtained with today’s multisection scanners.

In ECG-gated scans, an irregular R–R interval may cause data inconsistencies that are not recognized by the reconstruction algorithm, resulting in stair-step artifacts through the heart (Fig. 35-8).

FIGURE 35-8 Stair-step artifacts caused by heart rate changes during the scan appear as horizontal lines extending through the heart in the three-dimensional (A) and multiplanar reformatted (B) images. On curved multiplanar reformation (C), there is a blurred segment (between arrows) representing a slab that was imaged during an arrhythmia. In another patient (D), with a milder change in heart rate during the scan, the segment (between arrows) is displaced but not blurred.

Administration of β blockers to lower the heart rate and to try to stabilize it may prevent arrhythmia during the scan. Some manufacturers provide software for R-tag correction in case an arrhythmia has occurred (Fig. 35-9). Looking for a quiet phase of the cardiac cycle with minimal motion may help minimize cardiac motion–induced artifacts. If prospective gating with sequential (axial) acquisition mode is used, heart rate changes during the scan cause stair-step artifacts through the volume (Fig. 35-10).

FIGURE 35-9 A, On the right, there is blurring of the proximal right coronary artery (white arrows) due to premature beats (black arrows). B, After deletion of the tagging of the premature beats, the visualization of the proximal segment of the right coronary artery is improved, with obvious occlusion (white arrows).

FIGURE 35-10 Stair-step artifacts throughout the volume (A, volume rendering; B, multiplanar reformation) are caused by heart rate changes during a prospectively gated sequential (axial) acquisition for calcium scoring.

Noise-Induced Artifacts

The number of photons striking the detector directly influences the noise. More photons means less noise and better image quality. Increased noise (as a result of inappropriate choice of radiation parameters, thin-slice reconstruction, and large patients) can cause streak artifacts and a grainy appearance to the image. Optimization of scan parameters may reduce image noise as well as add several thin sections together into a thicker slab.

Poor Vessel Enhancement

Poor contrast enhancement within the lumen of the coronary arteries impairs the ability to interpret the study because of poor contrast-to-noise ratio. Technical error in the administration of the contrast agent as well as patient-related factors may cause such poor enhancement. Technical errors are usually operator dependent and include extravasation of contrast material, inadequate volume or injection rate, and improper timing of the scan. A well-prepared operator can easily avoid such errors by updating scan protocols for different clinical settings and by using high-density contrast material with appropriate injection protocols. Use of automatic scan triggering, with accurate scan delay and scan parameters, may help avoid human errors. Patient-related factors include Valsalva maneuver, heart failure, and obesity. Instructing the patient before the scan as to the importance of cooperation always helps.

Incomplete Coverage

Incorrect planning of the scan range, due to either a technical error on the part of the operator or different breath-hold depth during the scan, causes incomplete coverage of the heart with missing data. This can be avoided by instructing the operator, using safety margins in planning the scan, and rehearsing the breath-hold with the patient.

Image Interpretation

Postprocessing

Images can be reconstructed during different cardiac phases, allowing retrospective selection of the phase with the least motion artifacts.

For easier evaluation of the large amount of CT data and the complex anatomy of the coronary arteries, several postprocessing techniques are available. Image postprocessing requires modification of three-dimensional data to derive additional information or to hide unwanted information. The isotropic sub-millimeter voxel, available in all advanced scanners, improves the diagnostic quality of rendered images. Analysis of rendered images as well as of the axial slices has become a crucial component of cardiac CTA interpretation.

Axial Images

Axial images are the basic outcome images from a helical scan and include all the information acquired. In cardiac CTA, axial images should be reviewed in all cases and serve as a reference to confirm any pathologic change found or suspected on rendered images.

Multiplanar Reformation

In multiplanar reformation (MPR), a plane is defined within the three-dimensional volume of the scan, and only the data in this plane are displayed. MPR can be performed in either a straight plane or a curved plane (along a vessel’s “centerline”). In performing MPR, the thickness of the selection is set to be as thin as the collimation allows. When a greater thickness is selected, a slab MPR is created, and it is usually rendered with maximum intensity projection (see later). MPR can be created in any plane; it is very fast and easy to use and provides images containing all available information (all Hounsfield unit values are retained). Curved MPR with cross-sectional images is the best method for stenosis assessment, particularly in patients with calcified lesions or stents (Fig. 35-11). It is possible to rotate the image around the centerline and to view any plaque from different rotational angles to differentiate between eccentric and concentric plaques. A major disadvantage of MPR is the dependence on manual orientation of the planes that may cause false-positive or false-negative interpretation of stenoses. Interactive viewing of these types of images from multiple viewing angles is therefore required. Furthermore, for each vessel branch, a separate MPR image is required, and only one branch can be displayed at a time. Advanced software is available to allow semiautomatic segmentation of the coronary arteries, to determine the vessel’s centerline, and to reconstruct curved MPR as well as cross-sectional images.

FIGURE 35-11 Curved multiplanar reformation with cross-sectional images is the best method for stenosis assessment, particularly in patients with calcified lesions or stents.

Maximum Intensity Projection

In maximum intensity projection, only the highest attenuation voxels are displayed. A thin-slab MPR is selected (usually in a plane parallel to a vessel of interest) from which a maximum intensity projection image is rendered, and this slab can be moved through the volume. Maximum intensity projection images allow fast evaluation of vascular structures on CTA studies as initial screening of the luminal integrity of the vessel. It is easy to look for irregularities in the artery, with negative or positive remodeling. Because maximum intensity projection images demonstrate only part of the available data, small soft plaques may be obscured, especially with thicker slabs. We have found that a 4- to 6-mm slab is best for initial navigation and evaluation of the coronary tree. With dense material (such as calcifications and stents), this method is unsuitable because it may appear enlarged or thickened with overestimation of stenoses.

Volume Rendering

Volume rendering consists of all pixels above a certain threshold and therefore represents only part of the information. Instead of gray values, color can be attributed to the voxels according to their CT number. Volume rendering provides information about three-dimensional anatomy and orientation in relation to other structures. It allows comprehensive overview of lesion location and myocardial tissue at risk. Volume rendering is operator dependent and should not be used for stenosis evaluation, especially in the presence of dense materials (such as calcifications and stents).

Stenosis Assessment

The rapid evolution of multidetector CT (MDCT) technology has revolutionized cardiac CT, previously limited by cardiac motion, slow acquisition times, and insufficient resolution. Image quality and diagnostic performance have greatly improved with recent technical advances. Many studies have compared MDCT (4-,16-, and 64-slice scanners) with invasive coronary angiography in the assessment of coronary artery narrowing.

The percentage of unassessable segments was more than 30% on a 4-slice scanner ⁴^,⁵ and around 22% to 29% with 16-slice scanners.⁶^,⁷ One study stated that if these unassessable segments were excluded or considered negative, 25% of patients with a significant stenosis would have been missed.⁷ With 64-MDCT, only 3% to 11% of coronary artery segments still cannot be evaluated.⁸^–¹¹

Sensitivities and specificities for detection of significant (>50%) coronary artery stenoses based on segmental analysis with 64-MDCT (with conventional x-ray coronary angiography as the standard of reference) have been found to be good to excellent, in the range of 76% to 99% and 95% to 97%, respectively.¹¹^–¹⁵ However, these study outcomes are difficult to compare, mainly because of the different selection of patients and the prevalence of significant coronary artery disease. Moreover, these study results should be interpreted with care because of exclusion of unassessable segments from analysis as well as studies with reduced image quality. In an unselected population of patients referred for coronary catheterization with a high prevalence of risk factors, we found high specificity (95%) and negative predictive value (95%) and moderate sensitivity (72%) for significant coronary narrowing with use of a 40-slice scanner.¹⁶ All studies found that cardiac CTA has a very high negative predictive value (95% or higher) for exclusion of significant coronary stenosis.

Accuracy for detection of stenoses depends highly on image quality and artifacts. False-positive and false-negative interpretations were attributed to image artifacts in 91% to 100% of cases,^10,¹²^–¹⁴ ^,17 ^,18 mainly because of the presence of calcifications. Less frequent causes were motion artifacts and obesity, resulting in a poor contrast-to-noise ratio.

As mentioned, calcified plaques are a major cause of overestimation of stenosis, mainly with the older generation of scanners (4- and 16-slice scanners). It was recommended to perform a non–contrast-enhanced scan before the cardiac CTA and to avoid scanning patients with high calcium scores (usually above 400 to 600). In our experience, using the 64-slice scanner, with improved temporal and spatial resolution, it is possible to cope with most calcified plaques. Curved MPR with cross-sectional images through the plaque is the method of choice (Fig. 35-12). If a lumen is visible adjacent to a calcification (regardless of its size), significant stenosis (>50%) can be excluded.

FIGURE 35-12 A, Curved multiplanar reformation with cross-sectional images through a calcified plaque in the proximal left anterior descending coronary artery. There is a visible lumen adjacent to the calcification (arrows); therefore, significant stenosis (>50%) can be excluded. B, In a different patient with multiple calcified plaques along the right coronary artery, it is possible to exclude significant stenosis on the basis of this curved multiplanar reformation.

Plaque Characterization

Acute coronary events are usually caused by rupture of atherosclerotic plaque (in most cases, nonobstructive plaque), platelet aggregation, and thrombosis with partial or complete occlusion of the arterial lumen. This is one of the reasons that many patients do not have symptoms before their first coronary event. When individuals at increased risk for acute coronary events are identified while still asymptomatic, initiation of preventive therapy, including antiplatelet, antihypertensive, and lipid-lowering medications as indicated, can substantially reduce the risk of coronary artery events. Traditionally, the classic risk factors have been used to identify individuals at risk (quantification of their risk by the Framingham score), but they have limited predictive accuracy. Cardiac CTA is a noninvasive imaging modality allowing identification, quantification, and to some extent characterization of the atherosclerotic process at a subclinical stage. Quantification of coronary calcium (calcium scoring) is an established method to estimate the coronary plaque burden, with a high predictive value for occurrence of future cardiac events in asymptomatic individuals, independently of the traditional risk factors.¹⁹ However, the calcium score (calcified plaques) represents only about 20% of the total plaque burden,²⁰ and it is the noncalcified component of the plaque that is considered less stable and prone to rupture.

In its early stages, atherosclerotic plaque is usually accompanied by an outward growth of the vessel (termed positive remodeling), indicating a large plaque volume without lumen narrowing. Invasive coronary angiography, the clinical “gold standard” for coronary artery imaging, allows visualization of the lumen only and therefore is not suitable for plaque imaging. Intravascular ultrasound (IVUS) is considered the method of choice for plaque visualization. However, this is a highly invasive and expensive modality, therefore unsuitable for routine use and risk stratification. Cardiac CTA is a noninvasive modality, widely available, that enables visualization of both vessel wall and lumen.

The ability of CT to identify and to characterize nonobstructive plaque composition (calcified, noncalcified, and mixed) has been demonstrated before,²¹^,²² and Achenbach and coworkers²³ demonstrated vessel wall remodeling in high-quality 16-slice scans. Schroeder and associates²⁴ studied 12 patients and found that “hypoechoic” plaques on IVUS had a lower mean CT attenuation (14 ± 26 HU) compared with “fibrotic” plaques (91 ± 21 HU) or “calcified” plaques (419 ± 194 HU). Leber and colleagues²⁵ found a mean density of 49 ± 22 HU for soft plaque, 91 ± 226 HU for intermediate (fibrous) plaque, and 391 ± 156 HU for calcified plaque, as defined by IVUS. Even though CT is able to detect a variety of densities within the plaque, in our experience and that of others, there is substantial overlap between those groups. Furthermore, contrast enhancement within the vessel lumen may affect plaque enhancement, leading to variability in readings for any given plaque.

Comparing the performance of the 64-slice scanner with IVUS for noncalcified plaque detection, Leber and colleagues ²⁶ found 84% sensitivity and 91% specificity, with good correlation between plaque and lumen area measured, especially in proximal segments of the coronary tree. Caussin and associates²⁷ found that 16-detector row CT can accurately assess certain vulnerable plaque characteristics, such as hypodense areas (representing lipid), eccentricity, arterial remodeling, and calcifications, in comparison to IVUS. However, detailed microanatomy and detection of inflammatory changes in unstable lesions are beyond the resolution of current CT scanners.

We compared the total coronary plaque burden (segmental analysis), visualized by cardiac CT, with the traditional risk factors in 97 patients who underwent invasive coronary angiography. We found that nonobstructive coronary plaques were better detected on cardiac CT (277 of 737) than on invasive coronary angiography (111 of 737). The overall plaque burden was significantly higher in patients with diabetes, hypertension, or longer history of coronary artery disease and correlated with the number of risk factors. Furthermore, we found that among symptomatic patients without evidence of calcifications in the coronary arteries, 20% had noncalcified plaques and 7% had significant stenosis on 64-slice cardiac CT.²⁸ Because the long-term prognostic significance of these findings is currently unknown, additional experience and especially follow-up are needed to determine whether the additional “effort” (radiation, contrast material, time) to detect noncalcified plaques is justified.

Reporting

A CTA report should be useful for the referring physician. It should begin with patient identification data and a brief clinical history as well as the indication for the current study. Next, a brief description of the procedural technique should be mentioned, including the type of scanner, type of contrast material and volume used, premedications (if given), and radiation dose.

Because the accuracy of cardiac CTA depends directly on image quality, it is very important to record the quality of the scan so that the referring physician will be able to assess its reliability. Address these questions: Is the scan assessable and reliable? What are the limitations of the scans? What are the reasons for artifacts? It is easy to answer these questions by a quick leaf through the axial slices or slab maximum intensity projection images of the best phase. If multiple phases are loaded, loop through all of them to choose the best phase and to assess integrity of the data. Changing into a “lung window” can help detect breathing artifacts. It is possible to look at the chest wall and cardiac margins in MPR or volume rendering techniques to assess for motion and stair-step artifacts. Assessment for cardiac motion is easier on an oblique MPR through the right coronary artery and circumflex artery, which are more susceptible to motion. If cardiac motion is detected, additional phases should be reconstructed, trying to find a motion-free phase (if the technologist is not trained to perform this assessment and to look for the best phase for the reader). Looking at the patient’s ECG can help detect arrhythmia or heart rate changes during the scan and prompt an R-tag correction if necessary and if software is available. Comment about each unassessable segment (including location and reason). In a coronal MPR, it is easy to assess contrast density and uniformity throughout the aorta and coronary tree as well as noise level within the scan.

Report the calcium score level (if performed) and its clinical significance for the patient’s gender and age. In the presence of a zero calcium score, even if the cardiac CTA quality is not optimal, the referring physician can be reassured of the very low likelihood of significant stenosis in a stable patient.

The next step is to report on each of the coronary artery segments visible (Fig. 35-13). Invasive angiography will usually visualize more distal segments than is possible with CTA, but vessels smaller than 1.5 mm are currently not suitable for an intervention (with stent placement or bypass grafting) anyway.

FIGURE 35-13 Soft (noncalcified) plaque causing significant stenosis in posterolateral branch of right coronary artery (arrows) in a patient with atypical chest pain and reversible perfusion defect in the lateral wall on thallium scan. A, Volume rendering. B, Curved multiplanar reformation. C, Enlarged section of the curve.

After the initial navigation with slab maximum intensity projection images to locate areas of suspected stenosis, we perform a curved MPR for each coronary artery and for each visible branch. Because CTA provides three-dimensional information, the view that shows the largest lumen is the correct one (Fig. 35-14). Invasive angiography, on the other hand, is a two-dimensional imaging modality that provides projections of the coronary tree, in which the tightest view is the correct answer. Therefore, it is important to look at every segment from different viewing angles (by rotating the MPR around the centerline) to appreciate correct lumen size. Plaques (especially calcified plaques) should be carefully evaluated in cross-sectional images to assess for a visible lumen adjacent to the calcification. Any abnormality should be confirmed in more than one phase to exclude artifacts.

FIGURE 35-14 Mixed plaque in the proximal left anterior descending coronary artery (arrows), shown by volume rendering (A), curved multiplanar reformation (B), and cross-sectional images (C), causing significant stenosis.

Considering the spatial resolution of current scanners in relation to the size of the coronary arteries, in addition to the fact that cardiac motion may be reduced but not eliminated completely, cardiac CT cannot be expected to accurately quantify stenosis. Instead, cardiac CT should be used as a “filter” for patients with suspected coronary stenosis before invasive procedures.

For each lesion, it is important to indicate location (ostial, proximal, mid or distal, relation to branches), composition (noncalcified, mixed, or calcified plaque), eccentric or concentric, and evidence of remodeling.

Instead of giving precise stenosis percentage, we prefer to categorize each lesion into groups according to suspected severity of stenosis and clinical relevance (Table 35-2). Our categories include normal (when the vessel is smooth and there is no evidence of plaque), nonsignificant or mild stenosis (when there is some irregularity or plaques causing up to 40% stenosis), borderline lesions (when a lesion is suspected to cause 40% to 60% stenosis), and significant stenosis (when a lesion is suspected to cause more than 70% stenosis up to total occlusion). Alternatively, quartile gradations can be used, such as 0% to 25%, 26% to 50%, 51% to 75%, and 76% to 100%. Remember to underestimate stenosis caused by a calcified plaque (because of the blooming effect), and as long as a lumen is visible, significant stenosis can safely be excluded. The term total occlusion should be used only when no contrast material is visible distal to a stenosis because it is not possible to differentiate on CTA between antegrade and retrograde filling of a segment (Fig. 35-15). It is important to comment on presence of calcifications within the occluded segment, indicating chronic or acute on chronic lesion, which may be more difficult to treat invasively. When a lesion is difficult to assess (because of artifacts or calcifications), use of a statement such as “stenosis cannot be excluded” may be reasonable.

TABLE 35-2 Optional Categories for CTA Stenosis Assessment with Clinical Impression and Recommended Next Steps