Table of Contents
- 1 Cat Scans and how they work
Cat Scans and how they work
The term “CAT scan” means Computerized Axial Tomography. As a rule, it is shortened to “CT scan”. The basic principle behind computed tomography is to acquire multiple views of an object over a range of angular orientations. These views are then manipulated by a computer program to form a coherent image. A CT image is typically called a slice, as it corresponds to a slice from a loaf of bread.
The earliest CT scanners provided one slice per scan. As the technology improved, the number of slices per scan increased making it possible to stack up the slices just like a loaf of thinly sliced raisin bread. A single slice shows a two-dimensional section through the raisin bread, looking at the face of the slice. This view reveals the size, position and the shape of each raisin in that slice. When you begin to stack up slices, you can begin to see the three dimensional position and shape of the raisins. This holistic approach to three dimensional analysis of the images has been expanded to 64 slices in some modern machines. We are used to speaking in terms of pixels (meaning picture elements) when dealing with ordinary digital imaging. However the term used when dealing with three dimensional picture elements is “voxels“.
A CT scanner consists of:
An x-ray source
Unlike the x-ray tubes used for taking intraoral radiographs, the CT must put out a continuous stream of x-rays instead of a simple pulse, and it must produce a thin, divergent, fan shaped beam.
But like any diagnostic x-ray, it must have the smallest possible focal point in order to avoid the penumbra effect. The smaller the focal point, the higher the resolution of the final image. (When we speak of resolution, we generally are really referring to the degree of “fuzziness” of the final image.)
Finally, the energy spectrum of the beam defines how well the x-rays can penetrate the subject, as well as their expected relative attenuation as they pass through materials of different density. Higher-energy X-rays penetrate more effectively than lower-energy ones, but are less sensitive to changes in material density and composition. Lower energy x-rays are more prone to attenuation by soft tissue, and since soft tissue is often the specific target of the CT scan, the beam contains a lot of low frequency x-ray photons.
A series of detectors
The sensors measure the extent to which the X-ray signal has been attenuated by the object. They are similar to the CCD’s used in dental radiology, except that the individual detectors are arranged in a one-dimensional line or arc instead of in a two dimensional array.
A rotating gantry
A mechanism upon which both the x-ray source and the detector are mounted. The source and the detector array are mounted on opposite sides of the gantry and maintain a fixed positional relationship. The center around which the gantry rotates is called the axis of rotation.
The Subject sits midway between the x-ray source and the one dimensional detector array. The subject is situated so that the slice desired by the clinician is centered on the axis of rotation of the gantry. The only part of the subject that is directly exposed to the beam is that portion immediately surrounding the slice to be imaged.
How the CT scanner sees a slice
Referring to the diagram above, the x-ray source projects a thin, fan shaped beam of x-rays through the slice of the subject that the clinician wants to image. The line detector is situated opposite the detector, on the other side of the subject. The x-ray source and the detector are mounted on the gantry in such a way that both the source and the detector remain opposite one another as they both revolve around the subject. The subject is seated so that the axis around which the source and detector revolve is aligned with the center of the desired slice.
The detector then begins to capture “images”. As the gantry rotates around the subject, the detector captures another image from a slightly different angle. Each image the detector captures is called a view. Any given view would look like a gray line with varying degrees of darkness along its length. The intensity of any given pixel is dependent on the degree that the x-ray beam is attenuated by the various structures that it encounters along its path from the source to the detector. (Attenuationis a general term that refers to any reduction in the strength of a signal.)
Going back to our sliced raisin bread analogy, one beam might encounter two raisins in its path, another three, and a third might encounter none. The beam encountering more raisins in its path would be more highly attenuated, and the resulting pixel would be lighter than the another pixel which might be darker because its beam encountered fewer or no raisins.
Any given linear view might not be too exciting, and can’t say much about the structure of the slice, but as the source and detector continue on their trajectory around the axis of rotation, a computer program compares different views taken from different angles, and adds them up to construct a complex two-dimensional image of the slice, which shows not only the position of each raisin in the slice, but also the minute shape and density of each raisin. In order to compose a complete image, the source and sensor need to complete only a half revolution around the subject.
The computer program that builds the two dimensional slice makes use of mathematical versions of the same principles that we learned on the shadow casting tricks page. A single intraoral x-ray image gives us a two-dimensional picture of objects perpendicular to the beam of the x-ray machine. By itself, a single image gives us no information about the third dimension, which is parallel to the beam.
Recall, however, that when two films of the same area are shot from two different angles, the use of perspective enables us to determine the relative buccal-lingual position of the objects on the film. Using the Clark Shift technique, we are able to deduce the relative position of the objects in the plane of the x-ray beam itself.
The two images above were taken from two different angles. The one on the left was taken straight on, while the one on the right was taken from a mesial angle. Notice that the two buccal roots have each moved distally with respect to the palatal root of the same tooth. When we shoot from the mesial the buccal roots appear to move distally, and palatal root moves mesially relative to the buccal roots.
This is how the computer compares all the views of the objects in the scan beam. Each tiny detector in the linear array records the exact density of the sum of all the objects that attenuate the beam on its way from the x-ray source to that detector. It then stores its information in the computer’s memory. It does this again and again for each of the hundreds of views it sees as the scanner moves around its trajectory.
After one complete revolution of the CT scanner, the computer uses a mathematical algorithm to compare each linear view with the other views in the series as the CT scan revolves around the subject. An algorithm is simply a process, or group of actions that is repeated over and over again. Each repetition is called an iteration, and after numerous iterations of the algorithm, an image is created.
For example, if, instead of shooting just two films in the Clark shift above, we were to shoot hundreds, increasing the angle of the beam slightly more mesially for each shot, and then used the images to make a motion picture film, the result would be a three dimensional “tour” around the teeth. In effect, we would be using a mechanical algorithm. Each image, followed by our visual interpretation would be an iteration. In the process of watching the film, we would be building up a mental picture of the three dimensional structure of the teeth and surrounding bone using only the information garnered from a series of two dimensional images. We do this using our innate understanding of the laws of perspective.
How the computer draws a slice
Computers do not have a built-in sense of the laws of perspective, however the computer in the CT scanner has a program with complex algorithms which use repetitive mathematical processes to create perspective. This algorithm converts the series of linear views of the detectors into an image of a slice of the subject in the plane of the x-ray beam.
The first step in creating the CT scan involves creating a series of two-dimensional images, one from each view, composed of straight lines drawn perpendicularly from each tiny detector in the linear array. The intensity of each line would represent the intensity of the x-ray beam reaching the corresponding detector. The angle of the lines on any given view are perpendicular to the sensor at that point in its trajectory around the subject. The intensity of each line depends upon the accumulated density of all the objects the beam encounters at any given point along its course from the x-ray source to the detector. The algorithm sees the intensity of each line as a probability that one or more objects lies along that path of the x-ray beam.
After all the images (views) are drawn, they are superimposed over each other. The lines in any given view will intersect with the lines from each of the other views, because each view was shot at a different angle. At each point where the lines intersect the probabilities are summed and averaged using complex math to give each pixel on the plane of the beam an intensity corresponding to the probable density that the beam encountered at that point. The more views from different angles that are incorporated into the analysis, the higher the probability that the intensity (shade) of that point on the image corresponds to the density of the material at that point in the plane of the beam.
As a simple example, lets look at a tubular loaf of raisin bread that we want to slice using a very simple CT scanner that sees only in black and white with no shades of gray. In other words, a pixel is either black or white, unlike a real CT scanner which shows thousands of shades of gray. In our example there are only two raisins in this slice, so it looks like this when we actually cut the loaf at that point:
We’ll start by looking at only two of the CT scanner’s internal views. The ones we’ll look at first are at right angles to each other. Remember that our simplified line detector sees only black or white pixels. For convenience, I have drawn the views in different, transparent colors for clarity. The only thing the computer knows at this point is that there is a 100 percent probability that there are objects in the slice wherever the colored lines are located, and zero probability in areas where there are no lines.
The next step in the algorithm is to combine the two views and mathematically calculate the probability of finding an object where the lines intersect:
Note that the lines intersect in 4 places. We know that there are only two raisins in the slice, but the computer does not. It knows only that there is a high probability that dense objects lie at these four points in the plane of the beam, and a zero probability elsewhere. Two of the black rectangles represent the real raisins, and the other two are “ghost images”. But which two are real? We solve the mystery with a third view from another angle:
Now the computer knows that there is a hundred percent probability that dense objects lie someplace along the two yellow lines, and also that there is a zero percent chance that any dense objects lie in areas not covered by the yellow lines. At this point we compare this image with the combined image above, and come up with the image to the left below.
Dense objects can exist only where all three color lines intersect, and this happens in only two areas. The superposition of all three images rules out the ghost images and makes the position of the real raisins definite. But it does something else as well. When we draw in the exact shape of the overlap of the lines, the rectangular shape of the images begins to soften and we begin to see the oblong, rounded shape of the raisins themselves. This is the result of only three views from three different angles (perspectives) using pixels that are only white or black. The CT scanner takes hundreds of views, from virtually all angles, and uses detectors that can discriminate black, white and 65,534 shades of gray. Dosing considerations
CT scans are most frequently done to image soft tissues. In order to image these tissues, the beam must contain a high concentration of low frequency, low energy photons. Because of this, much more of the x-ray beam is absorbed by the patient than would be the case with higher energy beams like the ones in imaging devices built primarily to image hard tissues. In addition to the dose delivered directly to the tissues of interest, there is quite a bit of scatter from these low energy x-ray photons. This increases the effective dose to remote tissues. The effective dose delivered by modern CT scanners is quite high, about 1.5 mSv for a head CT scan. Compare that with the 0.050 mSv delivered by a full series of intra oral films (19 films). Some health facilities advertise full body scans as a preventive measure, however, it can be estimated that the radiation exposure from a full body scan is the same as standing 2.4 km away from the WWII atomic bomb blasts in Japan. In a comprehensive survey in the United Kingdom, CT scans constituted 7% of all radiologic examinations, but contributed 47% of the total collective dose from medical X-ray examinations in 2000/2001.