Computed Tomography

Computed Tomography

In radiography or fluoroscopy, one is creating a shadow pic-ture or a projection of the attenuation properties of the human body onto a plane. Thus, each ray from the source to a given point on the film, such as ray FP in Figure 2-2, con-veys information about the sum of the attenuation along a line in the body; that is, anatomic structures are piled on top of each other and flattened into the radiographic image. In an attempt to give a different perspective, one may obtain projections from two different directions (eg, a lateral and an anteroposterior radiograph), so that the structures that are piled on top of each other differ in each projection. In the late 1960s a British engineer, Geoffrey Hounsfield, con-cluded that if one obtained projection data from a sufficient number of different angles, one could reconstruct the atten-uation properties of each volume element in the body and display these as a cross-sectional image. This required the computational power of a computer to accomplish, and the basic idea is illustrated in Figure 2-4. The x-rays from a source are detected by a series of individual detectors (rather than film) after penetrating the body, and each detector de-fines a ray from the source through the body, thereby creat-ing a projection. The width of the x-ray beam in the dimension perpendicular to the page is only about 10 mm; hence, only one slice of the body in the longitudinal direc-tion is imaged at a time.

The x-ray tube and the detector bank are rotated 360 de-grees about the patient to obtain, for example, 720 projec-tions at 0.5-degree intervals. The computer is then able to reconstruct a cross-sectional image of the slice of the body by dividing the slice into an imaginary matrix. In a matrix of 512 x 512 pixels in the transverse plane, each pixel represents an area of about 0.5 x 0.5 mm in a 25-cm diameter body. The computer assigns a numerical value to each pixel, which rep-resents the amount of attenuation contributed by the volume element of the body represented by that pixel, and these numbers are converted into a gray-scale image for viewing. In an axial scan series, after one slice is completed, the patient is advanced via a motorized couch by 10 mm in order to image the adjacent slice, and up to 30 slices (images) may be done to reconstruct the anatomy over a 30-cm length of the patient. Newer scanners, called helical (or spiral) CT scanners, use a continuous advance of the patient through the scan beam rather than the stepping couch motion utilized in axial scans, and axial slices are reconstructed by interpolation of data into the slice from a complete rotation. Multislice helical scanners with subsecond rotation times have been developed that collect data for reconstruction of several slices in each rotation; thus, a 30-cm length of patient anatomy can be im-aged in 15 seconds or less.

Figure 2-4. Computed tomographic (CT) scanning geometry. A single projection of the head is illustrated.