Computed Tomography and Computed Tomographic Arteriography in the Evaluation of Abdominal Aortic Aneurysms

Data Sets

Two-dimensional data (a slice showing an x and y plane) are constructed of many small dots termed pixels. Each pixel is identified by a position in the x -direction and the y -direction, as well as a score in the gray-scale color spectrum, termed Hounsfield units. The Hounsfield unit (HU) scale is bounded on one end by air (and represented by black or negative numbers) and on the other end by bone (represented by white and positive numbers), with water being neutral and defined as 0 HU. Each pixel has a resolution in the x and y plane that can be considered the ability to discriminate between two points in addition to the HU value that is ascribed. Typically, the x–y resolution of a CT scan is 0.4–0.6 mm at best. However, when data are obtained in a helical format by conventional MDCT, each pixel is by definition extended in the z -plane direction, so it represents a volume rather planar unit, and the volume is termed a voxel. Each voxel has an x, y, and z coordinate and a resolution in the xy plane as well as the z -plane. When the resolution in the xy plane is equal to the z -plane resolution, the voxel is a cube, and this is called isotropic resolution. Postprocessing of reconstructed imaging is most accurate when a scan is obtained and reconstructed as close to isotropic resolution as possible.

If one desires to analyze blood (HU +30–45) as opposed to other structures largely composed of water, it makes sense to try to create a larger HU gradient between blood and everything else. This is accomplished by administering intravenous contrast (HU +130). Ultimately, the ability to discriminate between two voxels depends upon the spatial resolution and the gradient between neighboring voxels based on Hounsfield units. Most studies are done in phases, where the anatomy in question is scanned both before and after the administration of contrast. The precontrast phase allows assessment of the vessel wall and is used to differentiate intramural hematomas from inflammation as well as assessment of calcification patterns. The arterial, or contrast, phase of the study best depicts the lumen of vessels and the opacification of end-organs with contrast. Sometimes a delayed-phase study is obtained that demonstrates regions of the vasculature or organs that fill slowly with contrast.

Scan Parameters

The quantity and quality of energy that is generated by the emitter is termed the tube voltage and is measured in peak kilovoltage (kVp). A high kVp penetrates tissues better, but it has the effect of decreasing the gradient of Hounsfield units between voxels. Thus, an obese patient requires a high kVp (e.g., 140 kVp) simply to provide an image. However, the contrast between tissues appears attenuated. A lower kVp setting can be used in thinner patients or children and results in a lower radiation exposure and better contrast differences between tissues. The amount of radiation delivered by the x-ray tube per second is represented by milliampere–seconds (mAs). A higher mAs means cleaner (less noisy) images because more photons are reaching the detector to create the image. Yet again, as the mAs is increased, so is the radiation dose, so there is always a tradeoff. Most often kVp and mAs are linked and are dictated by the specific protocols, but modifications can be made to the acquisition parameters if indicated to optimize image quality.

Two other factors are important during MDCT studies. The first is collimation, which allows regulation of the thickness of the x-ray beam. The second relates to how fast the patient is transitioned through the gantry, termed table speed. Table speed, collimation, and the speed that the gantry rotates defines the pitch of the helical data that are acquired. If the table quickly advances a patient through a slowly spinning gantry, the scan will be fast, but the z -plane resolution will be poor. Alternatively, if the gantry speed is increased or the table speed is decreased, a tighter spiral of data will be obtained, improving the z -plane resolution. Similarly, if a thinner collimation is used, finer data cuts are acquired. In general a pitch less than one is optimal for later reconstructions, but the newer MDCT machines use alternative reconstruction algorithms that do not allow a simple calculation of the pitch.

Temporal resolution of scans is also important and relates to the dynamic nature of the vascular beds. Electrocardiographic (ECG) gating is a technique that allows the acquisition of data during a specific phase of the cardiac cycle (usually during diastole) or during multiple points during the cardiac cycle. Although technological advancements have minimized the need for slow heart rates during ECG-gated MDCT, it is important that patients are not tachycardiac. Irregular heart rates also make gating challenging. These scans often have higher radiation exposure and limited scan length, and thus they are rarely used for aortic studies distal to the ascending aorta. Multiphase studies can be acquired that depict data throughout the cardiac cycle (as many as 20 phases) and illustrate the four-dimensional anatomy, with time being the fourth dimension. Thus specific measurements can be made during peak systole or end diastole, establishing a range of diameters of the vessel over the cardiac cycle.

Reconstruction

The data acquired from an MDCT scan are reconstructed into two-dimensional axial images with a slice thickness (roughly the z -plane resolution). The thickness of the slices may be overlapped, so that the same point in three-dimensional (3-D) space on the patient is represented on two different slices. This method optimizes the spatial resolution of the study and helps to improve the accuracy of the 3-D reconstruction. Multiplanar reformatting (MPR) is the most common method used to display a CT scan today, and it typically consists of a panel that depicts the axial, coronal, and sagittal reconstructions. However, because the data exist in three dimensions, the angle of each planar reconstruction can be modified, allowing the clinician to interpret the specific relationships among various structures.

Maximum intensity projections (MIPs) are created by ascribing a minimum Hounsfield unit value to a region. Then, all the voxels that exceed the minimum value are grouped and are projected at a uniform Hounsfield unit value that is equal to the highest Hounsfield unit value within the grouping of points. This connects the high-intensity regions of contrasted-enhanced vessels in three dimensions, which can then be viewed from any projection. As vessels that pass back and forth between planes they can be viewed by scrolling through any one of the image sets. Alternatively, a curved MPR can be created and allows interpretation of a specific vascular bed. This is done by creating a reconstruction of the images along the center line of flow (CLF) of a vessel.

Usually MIP images define the borders of a vessel, allowing the geometric center to be calculated. The dots representing the vessel’s center in 3-D space are connected to form the CLF. The vessel can be projected along a curved CLF ( Figure 1 A) or a straightened CLF ( Figure 1 B), allowing length measurements to be obtained. The vessel’s diameter is also most accurately measured using CLF projections. An image is reconstructed that is perpendicular to the CLF, and the diameter of the vessel is measured from this image ( Figure 1 C). This obviates the need to obtain maximum and minimum diameters of vessels from images that likely cut obliquely through a vessel, which provides an inconsistent and less accurate way to measure the diameter of aneurysms or stenoses.