Image Quality

Factors affecting image quality and characteristics of CTA images include temporal and spatial resolution. Spatial resolution refers to the degree of blurring in the image and the ability to discriminate objects and structures of small size. Axial resolution within the scan plane can be improved by using a small field of view, larger matrix size, smaller focal spot, and smaller detectors. The demand for high spatial resolution to visualize the various coronary segments that course with decreasing diameter to the apex is high. One of the major goals of MDCT technology development has been to obtain similar spatial resolution in all directions, also expressed as isotropic spatial resolution.¹⁹ CTA requires high spatial resolution to enable accurate detection and interrogation of small arterial branch vessels. Both in-plane and through-plane spatial resolutions are important indices that require optimization to provide isotropic source data. Isotropic means that the spatial resolution is approximately equal in all planes, which is necessary to enable multiplanar image reconstruction after acquisition. Isotropic data avoid loss of spatial resolution in one plane and reduce partial volume effects after source data reconstruction.²⁰ The increased scanner detector numbers has allowed increased through-plane spatial resolution as the detector width is reduced with increased numbers (1 to 1.25 mm with 4-detector row CT, 0.5 to 0.625 mm with 64-detector row CT).²¹ The new MDCT scanners have not improved in-plane spatial resolution; this is determined by detector geometry and the convolution kernel (reconstruction algorithm).²⁰ Temporal resolution is defined as the required time for data acquisition per slice. It represents the length of the reconstruction window during each heart cycle, which is determined by the gantry speed.

Noise is the random fluctuation of pixel values in a region that receives the same radiation exposure of another. Noise is an important determinant of CT image quality and limits the visibility of low contrast structures. It is determined primarily by the number of photons used to make an image (quantum mottle). Increasing tube current and voltage can reduce it. It can also be reduced by increasing voxel size (decreasing matrix size, increasing field of view, or increasing slice thickness).

Contrast is the difference of the intensity of one area relative to another. Image contrast is the difference in the intensity of a lesion and that of the surrounding background. CT is superior to conventional radiography for detecting low contrast differences. CT contrast is the difference in the Hounsfield unit (HU) values between tissues. CT contrast increases as tube voltage (kV) decreases and is not affected by tube current (mAs). Adding a contrast medium such as iodine can increase contrast. The displayed image contrast is determined by the CT window (width and level).

Scanner Design

Since the first pencil beam EMI scanner in 1972, there have been numerous generations of CT scanners. The first four generations of scanners all had a single row of detectors (single slice) with evolving x-ray tube and detector configurations. Initially, for these generations, the image data were acquired one slice at a time. This involved scanning a slice and then moving the patient table to the next slice position and scanning again, otherwise known as step-and-shoot. The next advancement was the introduction of helical (spiral) CT, which involved continuous moving of the patient table through the CT gantry, as the tube rotated around the patient. The relationship between the patient and tube motion is called pitch, which is defined as table movement (mm) during each full rotation of the x-ray tube divided by the collimation width (mm). A faster pitch means thicker slices, reduced resolution, but lower scan time and lower patient dose. A pitch of greater than 1 will leave gaps between slabs, and a pitch of less than 1 will allow necessary overlap among slabs.

Electron-beam CT technology, also known as fifth-generation CT, provides excellent temporal resolution that allows freezing of cardiac motion through very short acquisition times. In electron-beam CT, the detectors are stationary. The x-ray source is fixed. It consists of a 210-degree ring of tungsten. This is bombarded by an electromagnetically focused beam of electrons fired from an x-ray gun. The patient is placed between the x-ray source and detector, obviating the need for moving any part of the scanner during the examination. Whereas electron-beam CT has better temporal resolution at present (50 ms), the spatial resolution is not nearly as good as that of MDCT because the collimators are too thick.

Detectors

The next major advancement was the development of MDCT. This incorporated the use of multiple rows of detectors to detect wider fan beams. This configuration makes more efficient use of x-ray tube output and covers a larger area in the patient’s z-axis (long axis) for every tube rotation. Unlike in conventional CT, the slice thickness of MDCT is determined by detector width and not by collimator thickness. The number of rows has increased at a rapid pace during recent years, covering 2, 4, 8, 16, 32, 64, 128, 256, and now 320 slices. The current driving force behind this rapid evolution is coronary artery disease.

Flying focal spot technology improves spatial resolution without decreasing the size of the detector elements by use of two overlapping x-ray beams without a corresponding increase in dose. Detector elements of 0.6 mm can be used to acquire 0.33-mm spatial resolution images.

Diagnostic performance in coronary CTA is primarily determined by temporal resolution, which is the required time for data acquisition per slice. The determinant of this is the gantry rotational speed. For the typical setup of a single tube and detector, half a gantry rotation is necessary to acquire the data for volume reconstruction, that is, temporal resolution is equal to half gantry rotation. The temporal resolution of a 64-slice CT scanner with a gantry rotation of 330 ms is 166 ms. For motion-free images to be obtained at any phase in the cardiac cycle, a temporal resolution of 10 ms is required. Achievement of such high temporal resolution is impossible with CT. As a result, cardiac CT phase reconstruction is centered on the quiescent or low-motion window in end-diastole. The postulated required temporal resolution for reliable cardiac imaging is in the range of 65 ms.²² One manufacturer has developed a 256-slice CT scanner (128 × 0.625) with a rotation time of 270 ms and temporal resolution of 135 ms. However, with automatic multisegmental reconstructions with voxel-based optimization, temporal resolutions up to 68 ms are achieved.

Other platforms use dual-source CT technology, which comprises two x-ray tubes and two corresponding detectors. The two acquisition systems are mounted on the rotating gantry, with an angular offset of 90 degrees. For cardiac imaging, a detector configuration of 64 × 0.6 mm is used, whereby two subsequent 64-slice readings with a flying focal spot are combined into two 128-slice projections, with isotropic spatial resolution of 0.33 mm. With the tube rotation time of 280 ms, data can be sampled over only 90 degrees of a gantry rotation (as opposed to 180 degrees with single-source systems), resulting in the industry’s fastest temporal resolution of 75 ms. The ultrafast tube rotation and table feed enable acquisition of the entire heart in 250 ms within a single diastolic phase without a breath-hold, resulting in reduced dose down to an unprecedented 1 mSv or less.

The newly developed 320-slice CT scanner with 0.5-mm detectors (16-cm coverage along the patient’s z-axis) images the body in a cylindrical fashion and can scan the entire heart prospectively, in one heartbeat without any table movement. This is expected to reduce the likelihood of both cardiac and respiratory motion artifacts.

Dual Energy

One of the more recent advances in CT has been the introduction of dual energy. In dual-energy mode, the two tubes emit x-ray spectra of different energy levels that are simultaneously registered in their corresponding detector array. At any given time point, this provides synchronous dual-energy image formation for the entire anatomy encompassed by the scan range. This enables differentiation of scanned tissues with different spectral properties, allowing the identification and subtraction of vessel wall calcification.

Retrospective Gating

Cardiovascular MDCT imaging is currently predominantly performed in the spiral (helical) mode; data are acquired by constant rotation of the x-ray tube/detector system throughout the entire cardiac cycle (continuous scanning). The data acquired are linked to the ECG tracing, allowing retrospective reconstruction of multiple cardiac phases when the study is complete. A specific phase within the R–R interval can be chosen to create a stack of images.

There are two techniques to retrospectively gate the scan to the ECG for image reconstruction. In one technique, the images are collected from one particular point in time in the cardiac cycle, which is defined as a percentage of the R–R interval. Alternatively, this point in time is defined at an absolute fixed time in milliseconds before or after the R–R interval. This latter method is better for irregular heart rhythms.

Retrospective gating allows faster coverage of the heart than does prospective triggering because images are reconstructed at every heartbeat. Continuous spiral acquisition allows overlapping of image sections and therefore permits 20% greater in-plane spatial resolution than that allowed by the collimator itself, resulting in a resolution of 0.6 mm for a 0.75-mm section and 0.4 mm for a 0.6-mm section.

Continuous acquisition throughout the cardiac cycle also allows retrospective reconstruction at different phases of the cardiac cycle. This permits selection of the best phases for each of the coronary arteries and their segments where there is least motion and best image quality. Retrospectively, individual heartbeats may be deleted, or the reconstruction interval for an individual beat can be shifted manually if there are arrhythmias or variable heart rates. Retrospective triggering is the preferred method of triggering for assessment of cardiac function and valve disease.

Prospective Gating

Prospective ECG triggering is a sequential scan in which data acquisition is prospectively triggered by the ECG signal in diastole. Data are collected only at a predefined cardiac phase, established by the operator before the acquisition.

With this technique, tube current is turned on only at a predefined point after the R wave (a constant cardiac phase). At all other times between each R–R interval, no radiation is emitted. This triggering method requires a regular heart rhythm; otherwise, the image created during each heartbeat will occur at a different part of the cardiac phases, resulting in artifacts. Prospective gating is most often used for calcium scoring. Prospective gating reduces the radiation dose by up to 10 times. However, this occurs at the expense of unavailability of systolic phases for additional image reconstruction (which may be needed if diastolic images are suboptimal). In addition, assessment of cardiac function is not possible.

Hsieh and coworkers²³ have developed a new approach for CTA referred to as prospectively gated axial. This technique uses a combined step-and-shoot axial data acquisition and an incrementally moving table with prospective adaptive ECG triggering. This method takes advantage of the large-volume coverage available with the 64-slice MDCT scanner that enables complete coverage of the heart in two or three steps. With this technique, the table is stationary during the image acquisition. It then moves to the next position for another scan initiated by the subsequent cardiac cycle. The result is very little overlap between the scans, significant (50% to 80%) reduction in radiation dose, and more robust and adaptive ECG gating. Earls and associates²⁴ reported an effective dose for the prospectively gated axial group (mean, 2.8 mSv) that was significantly lower than that for the retrospectively gated helical group (mean, 18.4 mSv). This represents a reduction in mean effective dose to the patient by up to 80% from the retrospectively gated helical to the prospectively gated axial approach.

ECG Dose Modulation

Dose modulation is used to help reduce the radiation exposure. During retrospective scanning, the tube current (mA) is turned on continuously during the examination. The tube current is dialed down by 80% during systole. Although the lower milliamperage may result in a suboptimal quality image, this may not be of clinical consequence. The reduced dose of radiation is therefore applied before 35% and after 80% of the R–R interval, which reduces the radiation dose by up to 40% of the usual dose. With most machines, dose modulation is the default mode and needs to be turned off if it is not desired. With increased heart rates, dose modulation cannot be used.

Reconstruction Techniques

Reconstruction of images for the evaluation of arteries uses thin slices with the smallest field of view (e.g., 18 to 20 cm) that encompasses only the heart. If the entire chest needs to be evaluated, a larger field of view is used with thicker slices. Additional data sets with a larger field of view that includes the entire chest should be reconstructed for the evaluation of extracardiac structures.

Kernel or reconstruction filters determine the balance between image resolution and smoothness (or signal-to-noise ratio). Smooth kernels are used for coronary CTA. Sharper kernels provide better definition of lumen borders and reduce blooming artifacts from high-density structures (e.g., stent) at the expense of higher image noise. The use of smooth kernels provides images with lower noise at the expense of possibly increased blooming of calcium and stents.

Fixed Scan Delay

Few authors currently use the method of setting a standard timed scan delay after the intravenous administration of contrast material to acquire images in the arterial phase.^35,36 By setting the same scan delay for every patient, one is assuming that exactly the same hemodynamic conditions exist in all patients. This method does not accommodate patients with any variation from the normal (i.e., low blood pressure, low cardiac outputs, hypovolemia, high outputs). We would not agree with employing this method of a fixed scan delay time.

Bolus Tracking

To more accurately determine the optimal scan delay after intravenous administration of contrast material in patients with variable hemodynamics, the technique of contrast bolus tracking can be employed. Used by many authors,^37,39 this is an efficient way to optimize arterial opacification. Initially, a single low-dose CT image is obtained, without administration of contrast material, at the level of the common carotid for the carotids, the ascending aorta for the coronary and thoracic aorta, and the celiac axis for the abdominal aorta and runoffs. A 10- to 15-mm² circular region of interest is placed inside the middle of the aortic lumen, and this will subsequently measure the Hounsfield units of the aortic lumen on subsequent scanning. At 8 to 10 seconds after intravenous administration of contrast material, serial low-dose monitoring CT scans are obtained at the same table position at 1-second intervals. When the region of interest detects a preset contrast enhancement level (usually 100 to 150 HU value), there is automatic triggering of the scanner to acquire images in the desired scan range. This time-efficient method ensures optimal arterial enhancement within the region of interest, which can be moved to a different arterial location if desired. It enables lower contrast use and reduces scan-to-scan and patient-to-patient variability in arterial opacification.

Test Bolus

Another method to optimize arterial contrast opacification is that of a test bolus or timing bolus acquisition. Described by a number of authors,^38,40 this technique involves intravenous administration of a small bolus (20 to 30 mL) of contrast material followed by serial CT data acquisitions at one table position, usually at the levels described for bolus tracking. Images are acquired, after a scan delay of 8 to 10 seconds, every 1 to 2 seconds for a predetermined number of images (20 to 40) or until the CT technologist chooses to manually stop the acquisition after the contrast peak within the aorta. A time versus Hounsfield unit (attenuation) curve is then generated by placing a region of interest over the contrast-opacified aorta. The time taken to reach peak opacification is then used as the scan delay for the actual CTA and thus corresponds to the time taken for the contrast material to pass from the intravenous injection site of interest. This method is useful as it detects variable transit times between patients with different hemodynamic states and allows individualization of scan delays.

Adaptive Method

The runoff vessels in the symptomatic limb can be problematic with CTA as a result of proximal/in-flow stenosis or, alternatively, distal hyperemia at the site of arterial ulceration; these can cause arterial flow discrepancy between the two lower limbs. Thus, one limb will have a faster flow rate than the other, resulting in good contrast opacification of the arteries ipsilaterally but failed opacification contralaterally. CT fails at eliciting dynamic information. This problem may be somewhat reduced by use of an adaptive method of contrast detection that was described by Qanadli and associates.⁴¹ This method is similar to that of a test bolus technique in that a small 30-mL contrast bolus is administered intravenously, followed by two low-dose (20 mA, 120 kV) CT acquisitions. The first is at the level of the descending thoracic aorta (at vertebra level T12) acquired 20 seconds after the start of intravenous administration of the contrast material for 10 seconds; the second level is the popliteal arteries below the knee acquired 30 seconds after the start of the intravenous bolus (i.e., after the aortic level is finished). Three time-density curves are created, one from the aortic level and one from each of the popliteal artery levels. The time to peak contrast enhancement is then determined for each level (aortic time = T1; popliteal time = T2). If the right and left popliteal times are different, the symptomatic leg or the longer time is taken to be T2. The aortopopliteal transit time (Tt) is determined from T2 − T1. The CT gantry rotation time and table speed are then set to match Tt in order that the popliteal artery will be imaged at peak enhancement. T1 is the delay time between commencement of the usual intravenous contrast bolus and commencement of data acquisition.⁴¹ Laswed and colleagues⁴² applied this adaptive method clinically in patients with peripheral arterial disease, with DSA correlation, and determined that MDCTA has a sensitivity and specificity of 100% for arterial lesion detection on a per-patient basis. Analyzed on a per-segment basis, MDCTA had a sensitivity and specificity for lesion detection of 91% and 96%, respectively, in the below-knee arteries and 100% and 90%, respectively, in the distal pedal arteries. This method was found to be reproducible, had high image quality, avoided the problem of venous overlay, and resolved the issue of differential peripheral arterial opacification.⁴²

Contrast Media Concentration

A recent review attempted to determine the difference that contrast media iodine concentration makes to image quality of MDCTA in the peripheral vasculature.⁴³ All of the studies reviewed used iodine concentrations of 300 mg/mL and higher; whereas improved arterial enhancement and visualization were demonstrated with higher iodine concentrations, there was no clear evidence of a significant difference in diagnostic efficacy for the different iodine concentrations.⁴³ Iezzi and coworkers,⁴⁴ comparing iodine concentrations of 300 mg/mL with 400 mg/mL on a 4-detector scanner, also determined no significant differences in diagnostic ability of CTA for peripheral arterial disease. This could perhaps be explained by applying findings from the coronary CTA literature, whereby Becker and colleagues⁴⁵ determined that coronary artery attenuation levels of 250 to 300 HU were optimal for the evaluation of coronary artery disease because higher attenuation values may underestimate the amount of atherosclerosis owing to obscuration of vessel wall calcification. In our institution, we use a contrast media iodine concentration of 350 mg/mL.

Dual-Head Power Injectors

To achieve the desired arterial Hounsfield unit, a fast iodinated contrast injection with a tight arterial contrast bolus is necessary. Thus, patients will need to have a well-positioned, large-bore (18-gauge), intravenous cannula within the antecubital fossa. A dual-head power injector permits both contrast material and saline to be administered separately, concurrently, and sequentially. This means that two separate injection phases are possible, the first with 100% iodinated contrast material and the second with a 100% saline flush. The advantages of this method are that the arterial contrast enhancement is improved and prolonged, the contrast dose is reduced because most of the administered contrast material is within the arterial side, and the saline flush at the end clears dense contrast material from the superior vena cava to avoid streak artifact.^14,30,46 The value of a triphasic injection strategy with an extra second phase comprising a contrast-saline mixture could also potentially reduce the requirements for contrast material. The new multislice scanners in combination with dual-head injectors allow marked decrease in the amount of contrast material used for routine CTA. However, the total volume of contrast material administered for a routine abdominal CT study (usually 150 mL of 300 mg/mL concentration of contrast material) should not change if solid organs, such as liver, are evaluated in conjunction with CTA. Lowering of the total volume of contrast material may potentially reduce the sensitivity of lesion detection.^12,32

Many authors have reported the use of gadolinium chelates for CTA in patients who have diminished renal function.^47–50 Although gadolinium is radiodense and may be used as a contrast agent with radiography, many gadolinium-based products have higher osmolality than iodine-based contrast media and are therefore potentially more nephrotoxic. More important, there have been several reports of nephrogenic systemic fibrosis leading to serious physical disability in patients with end-stage renal disease receiving gadolinium-containing contrast agents.^51,52

Cerebral CTA

Intracranial vascular disease can now be identified and characterized noninvasively by visualization of the blood vessels with contiguous thin slices and creation of isotropic maximum intensity projection (MIP) images and three-dimensional volume rendered reformations. The newer generation of MDCT scanners, such as the 320-slice CT scanner, allows dynamic entire-volume imaging, producing CT-DSA in all dimensions (two-, three-, and four-dimensional). The time-resolved data sets can be used to perform three-dimensional perfusion with quantification. This technique is not likely to completely replace conventional diagnostic DSA as it has inferior spatial and temporal resolution.

The scanning volume of CTA is determined on the basis of the location of vascular lesions or suspected vascular lesions on CT, MRI, or magnetic resonance angiography. If the lesions are suspected in the supratentorial or circle of Willis region, the scan volume begins at the level of the sellar floor and extends cranially. If the lesions are multiple or unknown, the scan volume can be extended caudally to the level of the foramen magnum. If required, the extracranial vasculature (carotids and aortic arch) can be evaluated by extending the scan volume farther caudal to the level of the aortic arch.

Automated bolus tracking is the preferred technique to reliably obtain the optimal arterial phase by placing a region of interest on the internal carotid artery close to the skull base. An injection volume of 75 mL of a contrast agent with high iodine concentration (350 mg/mL) at a flow rate of 5 mL/sec is typically used. Images are acquired in the caudal to cranial direction with a slice thickness of 1 mm, slice interval of 0.5 mm, and 0.6 mm of collimation (Fig. 80-1).

FIGURE 80-1 Axial MIP image (A) and volume rendered image (B) of CTA of the intracranial vessels demonstrating normal anatomy.

Carotid CTA

In the United States, stroke is the leading cause of adult disability and the third leading cause of death; only heart disease and cancer cause more deaths annually. About 80% of strokes are ischemic strokes. Ischemic cerebrovascular events are often related to atherosclerotic narrowing at the carotid bifurcation. Conventional DSA is the reference standard for the evaluation of carotid artery disease.^53,54 The advantage of noninvasive CTA is the reduction in cost and procedural risks. With conventional DSA, several studies have demonstrated up to a 4% risk of transient ischemic attack or minor stroke, a 1% risk of major stroke, and a small (<1%) risk of death.⁵⁵ Plaque composition can be evaluated by CTA, which is an important indicator of plaque stability and helps with the assessment of risk factors for thromboembolic events. Noninvasive CTA provides the necessary information required before carotid endarterectomy and endovascular therapies.