Dr. Thornton attended The Juilliard School, earning the BM, MM, and DMA in Piano Performance. He earned his medical degree from the University of Pittsburgh and is currently a third-year resident in the Department of Radiology at University of California, San Francisco. Following residency, he will complete a fellowship in Vascular and Interventional Radiology at UCSF.

Coronary artery disease is the leading cause of morbidity and mortality in the United States. Both electron beam computed tomography (CT) and multidetector row CT can produce diagnostic quality images of the coronary arteries following intravenous injection of contrast material. Speed and appropriate timing of image acquisition are key determinants of successful technique, and familiarity with the electromechanical relationship of the electrocardiogram (ECG) tracing to the cardiac cycle is fundamental to understanding the protocol. Images may be used to diagnose coronary artery atherosclerotic disease and pathologic states of coronary anatomy, to document the patency of bypass grafts and intraluminal stents, and to derive data regarding the functional status of the heart. This paper will review the technologic and physiologic principles fundamental to imaging coronary arteries with CT.

Heart disease is the leading cause of mortality in the United States, accounting for about 460,000 deaths annually. The most recent prevalence data from 1998 indicate that 12.4 million Americans suffer some manifestation of coronary artery disease. Every day, 600 Americans die from coronary disease outside of the hospital, and 20% to 25% of first coronary events are sudden cardiac deaths. Among those who die suddenly, most often, two or more major coronary arteries are narrowed by atherosclerosis. ¹

Noninvasive imaging studies play an important role in evaluation of the patient with suspected coronary artery disease. Nuclear scintigraphy is used to assess for abnormalities of myocardial perfusion, and gated acquisitions permit quantitative determination of left ventricular ejection fraction and analysis of wall motion. Echocardiography evaluates cardiac morphology and function but cannot reliably demonstrate coronary arterial morphology. Magnetic resonance (MR) imaging reveals cardiac anatomy in multiple planes and can be used to assess cardiac function, myocardial perfusion, and viability. This technique continues to evolve. Scoring of coronary artery calcium content by computed tomography (CT) is a relatively new method that possesses the ability to estimate the total coronary arterial plaque burden, but cannot specifically assess the severity of obstructive coronary arterial disease.

Cardiac catheterization is the most commonly used method for direct examination of the coronary arterial lumen. More than 1 million coronary angiograms are performed each year, at an estimated cost of more than $3 billion. Furthermore, transcatheter angiography is an invasive examination with a complication rate of 1% to 2% and an associated mortality of 0.15%. Nearly one in five cardiac catheterizations demonstrates normal coronary arteries or luminal stenoses of <50%. ^2,3 Therefore, a less expensive and less invasive method for evaluation of the coronary arteries has long been sought.

Until recently, electron beam CT (EBCT) was the only CT method capable of the fast exposure time required to image coronary arteries effectively. Multidetector row helical CT (MDCT) technology employs faster tube rotation times than older single-detector helical CT systems, and has the ability to acquire multiple thinner sections simultaneously, thus making helical CT evaluation of the coronary artery lumen now possible.

Electron beam computed tomography

Electron beam CT was developed by Imatron, Inc. (South San Francisco, CA) in 1983, based on research performed at University of California at San Francisco by Douglas Boyd and coworkers. Conceptualized to permit imaging of the cardiovascular system, the scanner's configuration has fundamental differences from helical CT design (figure 1). Unlike helical CT scanners, EBCT contains no X-ray tube. Instead, an electron gun discharges a 130 keV electron beam into a vacuum funnel. This beam is deflected by electromagnetic coils such that it strikes one of four semicircular tungsten targets, which lie in arcs beneath the patient. The resultant X-rays are collimated into a fan beam prior to penetrating the patient. After passing through the patient, the X-ray photons strike a semicircular detector array positioned above the patient, opposite the tungsten targets. The scanner's use of an electron gun rather than a moving, mechanical X-ray tube permits image acquisition in either a 50-msec or 100-msec mode, allowing excellent temporal resolution for cardiac imaging. Compared withcurrent conventional CT scanners, EBCT images have increased noise and may appear somewhat grainy. This and the higher equipment cost of EBCT may explain why the systems are not in widespread clinical use.

Multidetector row computed tomography

The introduction of multirow detector arrays to helical scanning technology in 1998 revolutionized CT imaging. Continuous rotation of the X-ray tube and the multidetector array during simultaneous table feed are features of the slip-ring technology characterizing this generation of scanners. Sub-second gantry rotation time and simultaneous acquisition of four or more helical slices offers fast volume coverage with high resolution. Many typical multidetector scanners have an effective temporal resolution of 300 msec, achieved by a 500-msec gantry rotation time and utilization of half-scan interpolation algorithms. Further improvements, not yet in clinical use, may reduce the effective exposure time to ¾ 125 msec. ⁴

When considering acquisition times for cardiac CT, note that a heart rate of 60 beats per minute (bpm) corresponds to cardiac cycle length of 1000 msec. Electron beam CT acquisitions are typically accomplished in 50 to 100 msec. The 300-msec temporal resolution achievable on many MDCT scanners, however, requires imaging during one quarter of the cardiac cycle at this heart rate. The current temporal resolution of MDCT is useful for patients with heart rates of ¾ 70 bpm. For heart rates significantly higher than 70 bpm, some investigators have administered beta-blockers prior to scanning to slow the heart rate to an acceptable level for MDCT imaging. ⁵

Successful depiction of coronary artery anatomy by CT angiography (CTA) is a demanding application. Near complete suppression of motion artifact caused by the beating heart is determined not only by the speed of image acquisition, but also by the precise synchronization of image acquisition with periods of minimal cardiac motion. In a process known as ECG gating, the ECG trace is used to synchronize the acquisition or reconstruction of image data to the cardiac cycle.

Known electromechanical relationships between the phase of the cardiac cycle and the ECG trace permit the portion of the cycle with the least motion to be chosen as the temporal window for imaging.

The cardiac cycle and coronary artery motion

The QRS complex heralds ventricular depolarization and the onset of ventricular systole (figure 2). Subsequent ventricular relaxation and repolarization occur during the T wave, resulting in diastole, the period of ventricular filling. Diastole can be divided into three portions: Periods of rapid ventricular filling, reduced ventricular filling, and atrial systole (which occurs on or around the P wave). The period of reduced ventricular filling, also known as diastasis of diastole or the "rest period," has been targeted as the optimal period for CT imaging because cardiac motion is minimal at this point. At higher heart rates, however, the rest period--the optimal imaging window--is disproportionately shortened compared with systole.

The distance between successive R waves (called the R-R interval) defines the length of a cardiac cycle-- the time between heart beats (figure 2). Because R wave amplitudes are large, they are convenient ECG signals for triggering acquisitions. ⁶ The time delay after the R wave, which serves as the trigger point for acquisition or reconstruction of coronary artery CTA images, can be expressed in either absolute or relative terms. An absolute time period, such as 450 msec, can be specified. Alternatively, the time delay can be prescribed as a percent length of the cardiac cycle, expressed as a percent of the R-R interval. For example, images acquired at 30% to 50% of the R-R interval correspond to mid-diastole, whereas the commonly used trigger of 70% to 80% of the R-R interval produces nearly end-diastolic images. ⁶

The coronary arteries, coursing along the surface of the beating heart within the epicardial fat, move with every heart beat. A recent angiographic study of coronary artery motion found considerable variation between patients in the range and pattern of coronary artery motion. ⁷ The length of the rest period for coronary arteries was strongly related to heart rate, as predicted from physiology. For heart rates >55 bpm, the rest period for the left and right coronary arteries asymptotically approached 66 msec. For heart rates <55 bpm, the rest period approached 333 msec for the left coronary artery and 200 msec for the right coronary artery. ⁷ The rest period was preceded by gradual slowing of coronary artery motion, and was followed by abrupt motion. The right coronary artery demonstrated both a greater excursion and velocity compared with the left coronary artery. Importantly, coronary arteries returned to the same location during the rest period of consecutive cardiac cycles. These findings support the presence of a window for near motionless coronary artery imaging and emphasize the challenge posed by faster heart rates. ⁷

Several recent studies report concordant results regarding the value of individual temporal windows for visualization of the right and left main coronary arteries. ^4-6,8 These studies report improved visualization of the right coronary artery (RCA) at points earlier in diastole than when the common default trigger of 70% to 80% of the R-R interval typically occurs. Kopp et al ⁵ reported that, although principal RCA segments could often not be evaluated at triggers of 60% to 70%, even branches of the RCA could be visualized at 40%. The left circumflex artery showed optimal quality in a trigger range of 50% to 60%, while the left anterior descending coronary artery was best imaged slightly later in diastole; between 60% and 80% of the R-R interval. Improved visualization of the RCA and left circumflex artery earlier in diastole is best explained by the effects of atrial systole. On or near the P wave, late in diastole, atrial systole imparts a ballistic effect on those vessels traversing the atrioventricular groove. Each of the coronary vessels may therefore be best evaluated at a unique portion of the cardiac cycle.

Prospective ECG gating

Electron beam CT coronary angiography and MDCT for calcium scoring utilize prospective gating, also known as ECG-triggered sequential scanning or the "step and shoot" method. Prior to scanning, using the ECG tracing, the operator specifies a delay interval after the R wave as the trigger for scan acquisition (figure 3). The selection of delay time is intended to image the coronary arteries during their period of least motion. Once selected, the delay interval is fixed for the examination, regardless of variation in the patient's heart rate or rhythm. Scanner software predicts the length of the R-R interval (and thus, the heart rate) based on the length of a variable number of preceding R-R intervals.

Advantages of prospective ECG triggering methods include a decreased radiation dose compared to retrospective methods, because scanning is confined strictly to those periods of the cardiac cycle that were prospectively chosen for image reconstruction. However, omitting portions of the cardiac cycle may result in suboptimal images because individual coronary arteries may be best evaluated in unique portions of the cardiac cycle. Current EBCT coronary angiography software addresses this issue by offering the acquisition of three prospectively determined scan intervals within each cardiac cycle. Other disadvantages of prospective ECG triggering include its sensitivity to arrhythmias and heart rate variability during the breath-hold period. Finally, the sequential scan technique does not provide continuous volumetric coverage of the heart. While this is acceptable for coronary artery calcium determinations, degradation in z-axis resolution is an impediment to high-quality CTA reconstructions.

Retrospective ECG gating

Coronary CTA performed with MDCT is gated retrospectively. A digitized ECG trace is continuously recorded during helical scanning following contrast injection. Slow table motion (pitch 1.5 to 3.0, depending on heart rate) is employed during multidetector helical scanning. Because the ECG is obtained simultaneously, postprocessing steps can match portions of the helical scan data set that coincide with any portion of the cardiac cycle. This permits retrospective reconstruction of data that is selective for any cardiac phase.

As in prospective gating, the portion of the cardiac cycle used for retrospective reconstruction is defined with reference to the R wave on the ECG tracing. Several different approaches to retrospective gating are technically possible (figure 4). The relative delay method designates the trigger time for image reconstruction as a fraction of the R-R interval after the previous R wave. Using this method, the delay time is calculated and applied for each individual cardiac cycle so that variations in heart rate have little or no effect on the quality of image reconstruction. The absolute delay method (also called the absolute forward method) defines the beginning of the reconstruction interval using a fixed time after the onset of the previous R wave. The absolute reverse method defines the start point for the reconstruction interval as a fixed time before the onset of the next R wave. The absolute reverse technique appears to be superior in cases of arrhythmia. ⁹

Advantages of retrospective gating include a decreased sensitivity to variable heart rate and rhythm. Continuous three-dimensional (3D) volumetric coverage of the heart permits reconstruction of data from any phase of the cardiac cycle. The principal disadvantage is a higher radiation dose compared with prospective techniques. The helical acquisition required for retrospective gating results in radiation throughout the entire cardiac cycle, even during those portions that are not useful for reconstructing images of the coronary arteries. Because each coronary artery may have an individual best phase for image reconstruction, it is estimated that only approximately 20% of the cardiac cycle could be potentially excluded in dose minimizing strategies. ⁸ Investigations are under way to synchronize reductions in tube output, with portions of the cardiac cycle not likely to be utilized for image reconstruction. ⁹

Scan parameters

Multidetector CT coronary artery angiograms require at least four detector channels operating at 1-mm or 1.25-mm collimation. Narrow collimation reduces partial volume effects. Pitches between 1.5 and 3 are selected by scanner software or by reference to a standard table based upon the patient's heart rate in order to assure continuous volume coverage. Using these parameters, the cardiac volume can be covered in about 30 seconds, well within the limits of a breath hold.

Contrast bolus technique

Optimal visualization of coronary arteries by CTA demands opacification of the arterial lumen with minimal venous opacification (figures 5 and 6). Circulation time is estimated in most protocols using a test bolus. Ten to 20 mL of contrast material is infused into a cubital venous line at 4 mL/sec and serial transaxial images are obtained through the ascending aorta every 1 to 2 seconds in order to determine the time to peak enhancement. Most investigators add 2 to 3 seconds to the measured circulation time to ensure adequate opacification of the coronary artery lumen at the time of scanning. For the diagnostic examination, 120 to 160 mL of contrast is injected at 3 to 5 mL/sec and imaging commences following the circulation interval with simultaneous recording of a digitized ECG.

Image postprocessing

Helical technique with 1-mm collimation and sub-millimeter reconstruction increments produces a near isotropic voxel, with unparalleled resolution in the z-axis, through which the coronary arteries run. Using dedicated software and workstations, the resulting 200 to 300 axial images can be reformatted for basic interpretation in approximately 5 minutes. Curved multiplanar formats, maximum intensity projections, volume rendering, and surface shaded displays can be quickly generated to optimally display coronary arterial morphology and spatial relationships (figures 7 to 13).

Radiation dose

Estimates for radiation dose for CT examination of the heart vary. Electron beam CT coronary calcium examinations deliver 0.8 to 0.9 mSv, while EBCT coronary angiography results in 1.7 mSv. Prospectively gated MDCT examinations for quantification of coronary artery calcium provide doses of 0.6 to 3.2 mSv. Reported doses from helical MDCT coronary angiography range from 2.4 to 8 mSv; Achenbach et al ⁶ report the effective radiation dose as 3.9 mSv in men and 5.8 mSv in women. Effective radiation dose reflects tissue-specific susceptibilities, and it is higher in women because they have breast tissue within the field. By comparison, a routine thoracic MDCT CT examination may result in a dose of 4 to 6 mSv, similar to the mean 5.6 mSv dose reported for diagnostic coronary angiography. Natural background radiation levels range 2 to 5 mSv per year. ^4,10-12

Image interpretation

The American Heart Association has adopted numeric segmental nomenclature for the coronary arteries. ¹³ Using this nomenclature in the description of stenotic lesions aids in communication with cardiologists and cardiovascular surgeons.

Abnormalities of the coronary artery wall deserve particular attention. A normal coronary artery wall is 0.1 mm thick and is not readily apparent on CTA. ¹⁴ Atherosclerosis may include both calcified and noncalcified components (figure 5). In fact, the most vulnerable plaques may have little or no associated calcium, being composed of a thin fibrous cap and a lipid-rich core. Such soft-tissue plaques may produce only modest lumen stenosis, but their rupture can lead to vessel thrombosis and occlusion, precipitating myocardial infarction. Detection of soft-tissue plaque, which can otherwise be evaluated only by intracoronary ultrasound, may prove valuable in risk stratification and monitoring of drug therapies.

Finally, caution must be exercised when evaluating coronary artery segments distal to sites of angioplasty or stent placement (figure 14). Opacification of the arterial lumen distal to a site of intervention may occur as the result of contrast passage through patent lumen or contrast passage to the distal lumen via collateral circulation. Time-attenuation curves, that compare peak enhancement and contrast transit times within a distal coronary arterial segment to controls, are routinely used in EBCT studies and should be equally useful for exams obtained on multidetector scanners. ¹⁵

Future directions

Patients suspected of stroke are often evaluated using a CT stroke protocol consisting of noncontrast brain CT, CT brain perfusion imaging, and CTA from the aortic arch to the circle of Willis. Similarly, a CT chest pain protocol may soon become a clinical reality. A noncontrast CT scan tailored for coronary calcium quantification could be followed by ECG-gated coronary angiography. Postprocessing steps enabling functional analysis have been described, permitting assessment of regional wall motion and left ventricular ejection fraction. ¹⁶

Summary

Multidetector CT scanners in current clinical use are routinely capable of retrospectively gated angiography of the coronary arteries for heart rates <70 bpm. Further studies are needed to clarify the appropriate use of beta-blockers in patients with faster heart rates. Bolus infusion of intravenous contrast material is followed by multidetector scanning of the heart at 1-mm or 1.25-mm collimation. Helical data is collected in a single breath hold with simultaneous digital recording of the ECG waveform. Data reconstruction is performed retrospectively in sub-millimeter increments with user-specified portions of the cardiac cycle chosen for optimal visualization of coronary arteries.

CTA of the coronary arteries is emerging from a boutique application restricted to academic referral centers to a powerful clinical tool for the evaluation of ischemic heart disease. The promise of sixteen or more detectors and further improvements in temporal resolution of scanners will make the technique even more robust. Attention to the requirements of the coronary artery CTA protocols will produce high-quality noninvasive coronary artery imaging and thereby crucial information to patients and referring clinicians.

Acknowledgments

The author would like to acknowledge and thank Michael B. Gotway, MD, Department of Radiology, University of California, San Francisco, and Harold I. Litt, MD, PhD, Department of Radiology, University of California, San Francisco, for their assistance in the preparation of this manuscript.