Most advances in cardiac CT, for example in coronary CT angiography (CTA), have focused on the development of protocols consistent with the rapid incremental improvements in the technology. One of the major technological advances has been the incorporation of multiple elements into the CT detector system, called Multi-Detector CT (MDCT). MDCT is synonymous with multi-slice CT. Data from each of the detectors is used to reconstruct an axial slice perpendicular to the long axis, or z-axis, of the patient. The width of the detectors determines the slice thickness and thus the ability to resolve small anatomic detail (spatial resolution) of the scanner. Thinner slices yield superior spatial resolution; however, comparing two scanners that produce the same number of slices, the scanner with thinner slices will have less z-axis (ie, craniocaudal) coverage per gantry rotation and thus will have a longer scan time. To date, the minimum detector width and largest number of detector rows is 0.5 mm and 320, respectively.1 This yields 16 cm (0.5 mm - 320) z-axis coverage per gantry rotation, and thus the entire heart can be imaged with data acquired over a single R-R interval.
Successful cardiac imaging by any modality relies on the ability of the hardware to produce motion-free images or, in other words, to image faster than the heart beats. Because it requires that the gantry be rotated around the patient, CT is inherently slower than digital subtraction angiography (DSA) where each frame corresponds to a single projection image. As described below, CT has recently become much faster and thus cardiac CT can now be routinely performed.
Temporal resolution is the metric that measures imaging speed. For a CT scanner with a single photon source, the temporal resolution is one half of the CT gantry rotation time. This is because image reconstruction requires CT data acquired from one half (180 degrees) of a complete gantry rotation. All manufacturers have gantry rotation times less than or equal to 350 milliseconds. Using this gantry rotation time as an example, an ECG-gated cardiac image can be reconstructed (using single-segment reconstruction, described below) with CT data acquired over 175 milliseconds of the cardiac cycle. Thus, the reconstructed images inherently display the average of the cardiac motion over the 175 milliseconds during which the data was acquired. This is how ECG gating enables cardiac CT. Without gating, cardiac images are nondiagnostic because the reconstruction “averages” the motion over the entire R-R interval, for example over 1000 milliseconds for a patient with a heart rate of 60 beats per minute.
There are additional strategies to improve temporal resolution. The first uses two independent x-ray CT sources and two independent (64-slice) detector systems built into the CT gantry. The second x-ray source is positioned 90 degrees from the first x-ray source, and the second detection system is positioned 90 degrees from the first detection system. With respect to temporal resolution, the practical consequence of this CT configuration is that 180 degrees of gantry rotation can be achieved in half the time (eg, 82.5 ms as opposed to 165 ms). This halves the temporal resolution (to 82.5 ms), and thus for this “dual-source” CT configuration, motion is averaged over only 82.5 ms. Another strategy uses both x-ray CT source-detector systems to sample the entire heart within a single R-R interval by rapidly moving the patient through the scanner.2 Both implementations of such systems have technical advantages and disadvantages that are beyond the scope of this chapter.
For single-source scanners, temporal resolution can be improved by adopting a so-called “multisegment” image reconstruction. The difference between single-segment and multisegment reconstruction is that in the former, 180 degrees of data are acquired from a single heartbeat, whereas multisegment reconstruction uses several heartbeats to obtain the one half gantry CT data. For example, in a two-segment reconstruction, two heartbeats are used to generate a single axial slice, and thus the temporal resolution is halved. Similarly, if four heat beats are used (four segment reconstruction), only 45 degrees of data are used from each heartbeat. This yields a fourfold reduction in the effective temporal resolution, making it theoretically possible to perform high spatial resolution cardiac CT in patients with a rapid (eg, >70 beats per minute) heart rate. However, because multiple heartbeats are used to fill the 180 degrees of gantry rotation necessary for the reconstruction, stable periodicity of the heart is essential. When beat-to-beat variations in heart rate occur, image quality is degraded significantly. In our experience, multisegment reconstruction works well in patients with high heart rates who are being studied for clinical indications where the highest image quality may not be required (studies of graft patency, pericardial calcification, etc.). For more demanding applications (eg, native coronary CT angiography) we still routinely employ beta-blockade for heart rates greater than 60 beats per minute.
Beta-Blockade for Heart Rate Control
As suggested above, beta blockade is an important component of most cardiac CT examinations. As the temporal resolution of cardiac CT improves, the dependence on lowering the heart rate will naturally be mitigated. However, the speed of all CT scanners is inferior to coronary catheterization, and thus beta-blockade is recommended for the large majority of patients in whom it is safe. In our experience, many surgical patients have standing orders for beta-blockers as part of their medical therapy, and image quality is excellent. When this is not the case, either oral and/or IV metoprolol is routinely administered.
ECG gating refers to the simultaneous acquisition of both the patient's electrocardiogram (ECG) tracing and CT data. By acquiring both pieces of information, CT images can be reconstructed using only a short temporal segment of the R-R interval. Each segment is named by its “phase” in the cardiac cycle; the common nomenclature is to name the percentage of a specific phase with respect to its position in the R-R interval. For example, reconstruction of 20 (equally spaced) phases would be named as 0%, 5%, 10%…95%. The period in which the heart has the least motion is usually (but not always) in mid diastole, near 75%. Thus, the CT exposure (and subsequent patient radiation level) can be lowered by limiting the exposure to a small part of the R-R interval where coronary motion is expected to be a minimum. This is termed “prospective” ECG gating because the reconstruction phase and width is determined prospectively. The disadvantage of this approach is that cine loops of the entire R-R interval cannot be reconstruction because there is no complete data set throughout the R-R interval. If this is desired, so-called “retrospective” ECG gating can be used, at the expense of high radiation levels.
For patients who require imaging of bypass grafts, it is important to note that periodic displacement of both saphenous vein grafts (SVG), radial grafts, and internal mammary artery (IMA) grafts is far less than the motion of the native coronary arteries. Thus, for these vessels, a single reconstruction at mid-diastole is usually sufficient (Fig. 6-2), and prospective ECG gating is routinely used. However, the benefits of lower radiation in this population are relatively moot, because the latent period for a radiation-induced malignancy is roughly a decade for blood tumors and significantly longer for solid tumors. Patients under consideration for repeat bypass surgery typically have a shorter life expectancy based on cardiac status. Thus, it is essential that the surgeon not only recognizes that motion has degraded image quality, but also realizes that additional reconstructions, and even repeat imaging, can and should be performed. If the entire course of the graft is not clear to the surgeon at a single cardiac phase, it is almost always the case that another phase will yield motion-free depiction of the graft segment that was poorly seen. At BWH, open communication between the radiologist and surgeon for every case has eliminated this pitfall and ensures that the maximum amount of imaging data is incorporated into presurgical planning.
ECG-gated CT images from a single reconstruction at middiastole for a patient scheduled for redo CABG. The patient is status post LIMA to LAD coronary bypass grafting. (A) Axial image demonstrates the LIMA graft coursing between two staples and adherent to the posterior table of the sternum. (B) Multiplanar reformatting is now performed routinely to detect and illustrate cases where repeat thoracotomy through the sternal incision is likely to damage a patent LIMA graft. An alternate surgical approach was required for this patient. (C) Selected image from a three-dimensional (3D) volume rendering again demonstrates the course of the graft. Volume rendering fully surveys the thoracic landmarks and is useful for spatial relationships and the communication of important findings
In cine CT, such as imaging the aortic valve over the entire R-R interval, images are acquired with retrospective ECG gating and subsequently reconstructed throughout the cardiac cycle and then played, in cine mode, to demonstrate function. Each individual image (Fig. 6-3) offers an outstanding assessment of the aortic valve and root structure. Cine CT can also be used to assess ventricular-wall motion. In comparison with magnetic resonance imaging (MRI), the gold standard for global- and regional-wall motion abnormalities, CT often has inferior temporal resolution. However, it is important to emphasize that cine CT does not require a separate image acquisition. The entire CT data set (coronary, valve, myocardium, pericardium) is acquired in a single breath hold; cine CT is simply part of the image postprocessing.
ECG-gated CT image through the left ventricle and the aortic valve in a patient status post aortic root repair. Note the pacemaker (right heart wires); magnetic resonance imaging (MRI) was contraindicated. The repair is well visualized and without complication, with only mild aortic valve calcification (cine images showed a tricuspid valve with no significant stenosis). This image also demonstrates a punctate calcified plaque along the superior course of the proximal left main coronary artery, without a significant stenosis.
For surgical patients, CT has the distinct advantage over MRI in that it is by far the best imaging modality to identify and quantify calcification. Also, the most common contraindications for cardiac CT (eg, impaired renal function as measured by glomerular filtration rate or alternatively by serum creatinine) differ from those for MRI (pacemaker), and thus CT can often be used for patients who cannot have MRI. Single heartbeat cardiac CT is now a clinical reality, with an entire cardiac acquisition in approximately one second.3 In addition to the fact that patient radiation is decreased, multiple scans can be performed with the same injection of iodinated contrast material, creating the opportunity for a host of additional studies (eg, myocardial perfusion)4,5 that are, at present, largely in the domain of cardiac MRI and nuclear cardiology.
Because ECG gating is required, ascending aorta and cardiac CT delivers more patient irradiation than CT of any other body part. Although details regarding cardiac CT dosimetry are beyond the scope of this chapter, discussions regarding CT dose must be based on sound principles. The radiation risk most commonly quoted relates to the probability that the CT scan will result in the development of a fatal radiation-induced neoplasm. Human data for radiation at this low level (the level delivered in ECG gated cardiac CT) is very sparse; all anecdotal reports support a long latency period as described above. For this reason, patients should be separated into two groups: those with a life expectancy of roughly 10 to 15 years or less, and those with a longer life expectancy. In the former group, the only dose consideration is whether the radiation could cause a skin burn (the only short-term complication of any consequence). X-ray skin burns are extremely uncommon, particularly in CT (even for ECG-gated studies), and typically result from multiple exams repeated at short-term intervals. Thus, for this subset of patients, radiation dose should not be a consideration in determining a modality for coronary imaging. For those patients for whom radiation is an important consideration, prospective ECG gating should be used. For cine evaluations in these younger patients, x-ray current modulation is a strategy to lower the radiation dose. The tube current (expressed as the mA) is modulated over the course of the cardiac cycle so that the desired (high) diagnostic current is delivered only in diastole. The patient dose is decreased because the tube current is reduced for the remainder of the cardiac cycle. Although current modulation is helpful in many cases (eg, pediatric patients), the decision to use it should be made after consultation between surgeon and radiologist because the potential drawbacks are significant. Most importantly, when current modulation is used, images reconstructed during phases with low tube current are relatively noisy because less tube current is used to generate them.
The scan time refers to the time required to complete the CT acquisition along the z-axis of the patient. As described above, better temporal resolution decreases the scan time, not only decreasing cardiac motion, but also enabling breath hold CT. This is important in cardiac CT because in comparison with nongated CT, ECG gating not only increases patient radiation but also increases the scan time.
In practical terms, a 64-slice ECG-gated cardiac CT scan (craniocaudal, or z-axis imaging over ~15 cm) can be performed in roughly 10 seconds, versus 20 to 25 seconds with a 16-slice scanner. One great benefit of wide area detector CT is faster (single heartbeat) scans. If this option is not available, increasing the thickness of the detectors increases the z-axis coverage per rotation and thus decreases the scan time. For example, in a patient who cannot perform the breath hold, using thicker detector widths (eg, 1mm thickness as opposed to 0.5mm thickness) will decrease the scan time by providing more z-axis coverage per rotation. However, routinely increasing the width of the detectors for cardiac applications is undesirable because it degrades the spatial resolution of the examination. In general, spatial resolution refers to the ability to differentiate small detail in an image. This is an essential component to coronary imaging because the diameter of the proximal coronary arteries are on the order or 3 to 4 mm. Substitution of 0.5mm reconstructed images with 1mm images thus impacts the ability to see detail that may be required for accurate diagnoses. Routine consultation between the surgeon and radiologist is essential to best understand and optimize the tradeoff between scan time and slice thickness. For example, imaging of the myocardium and ascending aorta almost never requires submillimeter slices, because the pathology is larger. Thus, for dyspneic patients who require only imaging of the ascending aorta, thicker slices are used to cut the scan time.
The scanning parameters that primarily determine the number of photons used to create a CT image are termed “mAs,” or milliamperes-seconds and “kV,” or kilovolts. The former represents the x-ray tube current; the latter refers to the voltage applied within the tube. For the surgeon, choosing the best numbers (typical values are 550 to 700 effective mAs, 120 kV) is far less important than understanding the fact that modern cardiac CT pushes the limits of technology, and thus creates tradeoffs with respect to the x-ray CT source. The source generates photons that are either attenuated by the patient or reach the detectors. When more photons reach the detectors, the image quality is higher because there is less noise. The decision to image with thinner slices (eg, 0.5mm as opposed to 1mm) means that fewer photons reach the detector; thus, thinner slices have more noise. This is especially important in obese patients because their increased body mass absorbs more photons than thin patients. For the same effective mAs and kV, images of obese patients can be dramatically degraded by greater image noise.
If there were no limit to the number of photons that an x-ray CT source could produce, the solution would be to simply increase the number of photons (and the radiation dose) until image noise was satisfactory. Unfortunately, because the x-ray CT tube heats excessively when pushed to its maximum, there is a limit to the number of photons that can be produced. This is why image noise becomes problematic with thin slice imaging of obese patients. When this is the case, consultation between surgeon and radiologist is important because diagnostic images can often be obtained by increasing the image thickness, scanning a smaller z-axis field of view (FOV), or both. The latter can be particularly useful if the exam can be tailored to the most important structure. Scanning a smaller z-axis means that more photons can be generated and used before the x-ray CT tube reaches its heat limit.
On the other hand, whenever possible, the z-axis FOV should be generous, as unexpected pathology can extend both cranially and caudally. For example, an ECG-gated cardiac and ascending aorta exam to evaluate extension of the intimal flap into the coronary arteries can reveal extension into the great vessels. Also, scanning must allow for variations in the FOV induced by breath holding. As a general rule, for scanning the native coronaries alone, the superior border of the FOV is set at the axial slice corresponding to the top of the carina. This is typically 2 to 3 cm superior to the origin of the left main. The inferior border should scan through the entire inferior wall of the heart and should include several slices of the liver to account for cardiac displacement during breath holding. For bypass graft imaging, the superior border of the field of view must include the subclavian arteries and the origin of both IMAs.
With the exception of scans performed solely for the assessment of cardiac and aortic calcification, CT examinations are performed with iodinated contrast material. Effective communication between surgeon and radiologist is important in optimizing the use of contrast material, particularly in the decision of whether to use a single or a dual injection system. A single system injects only contrast, whereas a dual system has two reservoirs to inject contrast followed by saline. Dual injection is essential for many applications, and it is routine in coronary imaging. The contrast and saline delivery are timed so that the left heart, aorta, and coronary arteries are enhanced with contrast while the right heart is filled with saline. The use and the timing of the saline are essential parts of the exam because artifacts that limit interpretation of the right coronary artery (RCA) will be induced if the right heart and central veins are densely enhanced with contrast (as opposed to saline). However, for patients in whom imaging requires enhancement of both the left and right heart (eg, assessment of both the mitral and tricuspid valves), saline cannot be used and less dense contrast material may be chosen to lessen potential artifacts.
Advances in technology such as submillimeter spatial resolution, up to 320 detector rows, and rapid gantry rotation times with dual x-ray CT source and detector systems have enabled ECG-gated CT to make a positive contribution to the care of cardiac surgery patients. The surgeon must recognize that cardiac protocols push the limit of technology, and thus CT of the heart is more complicated than a scan of any other body part. Consequently, routine and effective communication between the surgeon and the radiologist will result in the best possible patient outcomes.