Background

The major disadvantage of conventional radiography during the first 60 years after its discovery was superimposition of shadows. Stereo-imaging techniques provided an illusion of depth but didn't solve the problem. Beginning in 1921, work began to eliminate the superimposition problem with a technique initially termed body section radiography (BSR; Seeram 1985a). In 1935, Grossman introduced the term tomography, derived from the Greek word tomos, which means slice or section, and graphia meaning describing (Eiseberg 1992). The basis for this new technique was a coordinated motion between the x-ray tube and the image receptor. In the simplest motion, termed linear, the x-ray tube moved in an arch starting from the head end of the patient to the feet. At the same time, the image receptor moved, under the x-ray table, in the opposite direction (Figure 3.1). A plane within the patient parallel to the table was in focus, and all the anatomy above and below that plane was blurred and out of focus (Figure 3.2). This point at the center of the section, termed the fulcrum, constituted the pivot or plane of rotation of the x-ray tube and film (Seeram 1985b). To bring other planes into focus, the focal plane within the patient had to be raised or lowered. The thickness of the section was determined by several factors including the distance the x-ray tube traveled, termed amplitude, and the complexity of the x-ray tube movement. The greater the amplitude, the thinner the section visualized. Linear tomography provided the least complex motion and according to Hiss (1983), with an amplitude of 30°, a 2.12 mm section was demonstrated. If the angle were increased to 50°, the section thickness would be reduced to 1.19 mm. One of the disadvantages of linear tomography was that it produced streaks on the film. To eliminate the problem, more complex types of motion were developed. The thinnest slice, approximately 1.3 mm, was achieved with a motion term hypocycloidal, resembling "overlapping loops" and was employed for examinations of the inner ear (Figures 3.3A, 3.3B, and 3.3C).

Although the modality became quickly accepted, there were a number of disadvantages. Since multiple levels within the patient would need to be imaged, at least four or five radiographs, termed tomographs or tomograms, were obtained from a particular region. Because

Figure 3.1 In linear tomography, the x-ray tube (A1) begins at the head of the x-ray table, and the image receptor (B1) at the opposite end. By the end of the exposure, the x-ray tube (A2) and the image receptor (B2) have reversed their positions.
Figure 3.2 Above and below the axis of rotation, known as the fulcrum, is a region termed the section thickness that is in focus.

the x-ray source and image receptor were both moving, the area irradiated could not be as tightly restricted or collimated, as would be the case during a single exposure. Both these factors resulted in high radiation doses to the patient. In addition, the blurring of the anatomy above and below the focal plane made the images more difficult to interpret.

In 1970, Electric & Musical Industries Ltd (EMI), England, achieved a monumental milestone in medical imaging. The company's previous achievements included Enrico Caruos's first recording in 1902 and establishing the Abbey Road recording studio of the Beatles. Godfrey Hounsfield, a physicist and engineer working at EMI laboratories, first demonstrated the modality that he originally called computerized transverse axial scanning (Seeram 2001). Allan Cormack, a South Africa-born medical physicist at Tufts University, Massachusetts, had earlier formulated the mathematics necessary to reconstruct the images. The impact of their work was so revolutionary that both shared the 1979 Nobel Peace Prize for their contributions to medicine and science.

Figure 3.3A a lateral radiograph of the head of a cat following an injection of contrast media into the cerebral ventricles. Note the mandible (A), tympanic bulla (B), the olfactory bulb (c), and body (D) of the lateral ventricle.

Figure 3.3B A tomographic section of the cat's head in a lateral position acquired with a hypo-cycloidal motion. Note that the olfactory bulb (a), body of the lateral ventricle (B), third ventricle (c), cerebral aqueduct (D), and fourth ventricle (E) were free from superimposition by the skeletal components of the skull. However, because the olfactory bulb and body of the lateral ventricle were slightly beyond the focal plane, both structures are somewhat blurred.

Figure 3.3B A tomographic section of the cat's head in a lateral position acquired with a hypo-cycloidal motion. Note that the olfactory bulb (a), body of the lateral ventricle (B), third ventricle (c), cerebral aqueduct (D), and fourth ventricle (E) were free from superimposition by the skeletal components of the skull. However, because the olfactory bulb and body of the lateral ventricle were slightly beyond the focal plane, both structures are somewhat blurred.

Since the modality incorporated a coordinated motion between the x-ray source and the image receptor, the term tomography applied. Instead of film, the image receptor was replaced by a detector that transmitted x-ray attenuation data to a computer. The x-ray tube and detector traveled over a 360° arch within a doughnut-shaped structure called the gantry. The patient was positioned on a table that moved into the gantry. The computer analyzed the attenuation profiles collected from a number of angles around the patient, mathematically superimposed the profiles, and generated an image of the patient's anatomy. Since all the data were collected from the axial plane of the patient, the new modality was eventually called computed axial tomography or a CAT scan.

Figure 3.3C With the focal plane moved more laterally, the olfactory bulb (a) is more clearly defined. Air bubbles (B) in the contrast media within the body of the lateral ventricle indicate the structure was included in the section thickness.

Due to the limited computational capacity of computers at the time, the first clinical CT scanners were small and restricted to imaging of the head only. These early units were in use between 1974 and 1976 and rapidly changed medical imaging procedures. The impact within radiology was so dramatic that immense pressure was exerted on manufacturers to develop units that wouldn't be restricted to the head. Subsequently, "whole-body" systems became available by 1976, and CT became widely available by about 1980. There are now thousands of CT scanners, not only in the United States but worldwide.

The first CT scanner developed by Hounsfield in his EMI laboratory took several hours to acquire the raw data for a single scan or "slice," and took days to reconstruct a single 80 x 80 matrix image from this raw data. The latest multislice CT systems can collect up to 64 slices of data in about 350 ms, and reconstruct a 512 x 512 matrix image from millions of data points in less than a second. An entire chest can be scanned in 5-10 s using a modern multislice CT system.

A more thorough consideration of the development and history of the modality can be found in Bushong (Bushong 2004); however, the major advantages and disadvantages relative to our discussion will be considered here.

The advantages were tremendous. Because the images were reconstructed in an axial plane, superimposition was completely eliminated. Since the patient was in or close to the center of the x-ray tube-detector rotation, called the isocenter, magnification and distortion were eliminated, making it possible to take direct measurements from the images. Due to the inherent construction, the x-ray beam leaving the source was tightly collimated, minimizing scatter radiation so much that soft tissue structure could now be visualized. With CAT scans, it was not only possible to detect brain tissue within the skull but also to differentiate the gray matter from the white matter of the brain.

Although the aforementioned advantages were significant advancements in medical imaging, this new modality was also quantitative. Up to this point, radiography was qualitative. Although there were attempts to quantify the images using standards of known density, all were subjected to the variability inherent to film processing. A difference of a few degrees in the developer temperature, a variation in the concentration of developer, improper fixation or washing, to name a few, would result in different density values. In addition, superimposition of shadows, such as gas or feces in the intestine lying over the vertebrae, would also provide different density values. This new modality eliminated these problems.

To construct the axial image, the computer calculates what is termed the linear attenuation coefficient, or lac. Since x-ray attenuation is dependent on the penetrating power of the x-ray beam, determined by the kVp setting, the lac for each pixel or picture element of the image must be adjusted. For this process, the pixel lac value is standardized to the lac of distilled water at a specific kVp setting. This standardized pixel lac in turn is used to calculate the "CT number." Each type of tissue or material will attenuate or absorb radiation differently, allowing for a tissue characterization and, therefore, a unique CT number. With distilled water as a standard, the CT number of water is 0. Regardless of the manufacturer and kVp setting, the CT numbers for specific tissue types are approximately the same; for example, the CT number for white matter is 30 and that for gray matter is 38 (Wolbarst 1993).

Postprocessing also permits the contrast and density of the image to be adjusted through an operation termed windowing. The CT number of the tissue of interest is selected as the level, or center, of a particular window of values to be displayed. The window width

Figure 3.4A An axial section through the skull at the level of the third ventricle with a narrow window width (WW) of 100 provided visualization of cerebral structures.

indicates the range of CT numbers displayed as one of 256 shades of gray. Any CT numbers to the left of the window will appear "black" on the image and those to the right of the selected values will appear "white." A narrower window, for example, one with a window width (WW) of 100, is utilized to differentiate tissues of similar density, such as gray matter (CT # = 38) and white matter (CT # = 30) of the brain. With a window width of 100, each CT number is separated by 2.56 shades of gray. Since the tissues are separated by only 5 CT numbers, at WW = 100, there are 12.8 shades (2.56 x 5 = 12.8) separating the tissues, enabling easy differentiation (Figure 3.4A). Wider windows, for example, WW of 500, are employed to suppress differentiation of subtle differences in tissue density, such as "eliminating" brain tissue entirely to permit an evaluation of the cranial bones (Figure 3.4B).

Until the mid-1990s, all CAT data were collected by a process called slice-by-slice acquisition. After each axial slice was collected, the table on which the patient was lying would move cranially or forward for a prescribed distance. Another slice would then be collected, and the process would continue until all the desired anatomy was scanned. If sagittal or coronal sections were required, a postprocessing operation, called reformatting, would stack all individual slices previously collected and "cut" them into the desired plane. There are several disadvantages to this procedure, but only the one relevant to this discussion will be considered here. Since the time between individual slices might be several seconds, the patient or their organs would probably be in a different location. The reformatted images would not show a smooth continuity along the edges of organs or structures (Figure 3.5).

Figure 3.4B An axial section at the same level as that in Figure 3.4A, but with moderate WW of 500. Note that only the skeletal components of the calvarium were visualized.

Figure 3.5 The axial image a indicates the level (dotted line) that was used to produce the sagittal reconstruction.

ifll w

Figure 3.5 The axial image a indicates the level (dotted line) that was used to produce the sagittal reconstruction.

Figure 3.6 A three-dimensional reconstruction from a study done in the mid-1980s. note the irregular margin of the right orbit (A) created by the stacking of the axial slices.

This problem was also noticeable in cases where three-dimensional (3D) reconstruction was desired (Figure 3.6).

The problem was solved when a new approach to data collection was introduced. Due to a technical innovation, known as slip ring technology, the x-ray tube and detectors could continuously rotate around the patient as the table constantly moved forward. The result was that instead of individual slices, a volume of anatomy was acquired. Since the volume could be postprocessed into any plane, the modality became known as simply CT. The more accurate description for the new generation of scanners incorporating this technical advancement is spiral or helical CT. The introduction of helical CT has greatly advanced the progress of medical imaging. Scans of a region of anatomy, such as the chest, can be accomplished in "one breath-hold" of the patient. At the time of this writing, the equipment is capable of collecting the equivalent of 64 0.5-mm-thick slices during one scan. Since the units can have multiple detectors collecting data during an interval, they are also called multidetector or MDCT.

In addition, software has been developed to perform specialized postprocessing operations. Algorithms designed for specific anatomical regions of a living patient have eliminated the need for technologists to manipulate variable settings on the scanner. Manufacturers are anticipating the future needs of the medical imaging community, and are introducing not only software but also hardware updates more rapidly. The impetus is to reduce the time required to image a patient while acquiring the greatest amount of information. All this innovation comes at a high cost. A 64-slice CT scanner can cost $850,000-$950,000. Once the unit is purchased, the costs do not end. The accompanying service or maintenance contract, which also includes software updates, can add another up to $120,000 in costs per year. The continuous output of the x-ray tube during the MDCT scan generates a tremendous heat load for the tube to overcome. In a busy imaging department that operates 24 h a day, 7 days a week, the x-ray tube probably will not last a year. Without the service contract, the cost of a replacement tube could exceed $80,000.

Was this article helpful?

0 0

Post a comment