Computed Tomography (CT) images and their reconstruction
Computed Tomography (CT) or Computerized Axial Tomography (CT) is a diagnostic imaging technique that provides tomographic images (i.e., of slices of body layers) by drawing the various organs and tissues on the basis of their density, detected due to the attenuation of an X-ray beam that passes through the patient from different points of view.
Furthermore, CT images are digital, and they are the result of billions of numerical calculations performed by a computer, which converts the density of the tissues crossed by the X-rays into gray levels.
Brief history
CT is a typical example of an invention or discovery that two scientists or two study groups come up with independently. In the United States, around the first half of the 1960s, a South African physicist, Allan Cormack McLeod, published his works in which he described the equations for reconstructing images based on the attenuation of an X-ray beam passing through a body plane from different angles. Independently, in the UK, electrical engineer Godfrey Hounsfield, who worked at EMI, developed a system for processing and converting multiple X-ray beams passing through a body in the same geometric plane into images. In 1972, the first commercial model of Tomograph was put into production at the time produced by EMI. The introduction of CT had a very high echo both in medicine and among ordinary people.
In 1979, Cormack and Hounsfield were awarded the Nobel Prize for Medicine.
How Computed Tomography (CT) works
A more or less thin layer of the body is crossed by a highly collimated X-ray beam, produced by a tube that rotates around the patient, in a consensual way to detectors placed beyond the patient.
The data relating to the attenuation of the beam, obtained from the different "points of view," are sent to a computer which, through complex mathematical algorithms, reconstructs the images of the anatomical structures present in the considered layer. The layered visualization of the anatomical structures eliminates the problem of overlapping present in the radiographic examination.
Voxel
Each body layer is divided by the machine into elementary volume units (Volume x Element = Voxel). The beam attenuation is calculated for each individual voxel. The dimensions of the voxels depend on the collimation of the beam and on the number and size of the individual detectors.
Attenuation profiles
In CT, since the intensity emitted and that of the intensity measured by the detectors are known, it is possible to calculate the attenuation profile that the X-ray beam undergoes for each column of voxel crossed.
Rear Projection
The data relating to the attenuation of the beam for each individual voxel represent the projection of a given object interposed between the source and the X-ray detector. The rear-projection method is used to reconstruct the image of the object; for each point of view or projection, the relative projected image is "rear-projected." The rear projection process consists of repositioning, through complex algorithms, all the recorded images, and then going back to the original ones by going "by exclusion." What is reported is the attenuation coefficient, the computer projects what remains of the X-ray after it has passed through the body.
Filtering
In the rear projection processes, the two objects have a sort of tail along with the profiles due to the partial attenuation of the beam. This partial attenuation is responsible for a blur along with the profiles. To eliminate, or better, mitigate this effect, the signal sent to the detector is filtered beforehand, by means of particular mathematical algorithms (convolution filters), before being rear-projected. There are different filters that accentuate more or less the "cleaning" of the profiles (bone filters, soft tissue filters, etc.).
Staircase and Hounsfield Unit
The density of the tissues is expressed by a grayscale built on the basis of the units (or numbers) of Hounsfield (UH), so-called in honor of the inventor of the TC. The Hounsfield unit or TC number is a dimensionless value proportional to the density of the fabric. The UH refer to the density of the water, which by convention is equal to 0. Above and below this value, the densities of the different tissues of living matter are located. Most soft tissues and organic liquids have a density between +100 and -100 UH. In the CT exams, we speak of density, the organs and tissues may be hyperdense, hypodense, or isodense in relation to another organ or tissue or the reference density of water.
Computed Tomography (CT) imaging
Based on its attenuation profile, each voxel is assigned a Hounsfield number. This number represents the average attenuation of the corresponding volume of tissue examined. Subsequently, each voxel, with its number in UH, is assigned to the image matrix, which is usually formed by 512 x 512 pixels. So, at the end of the process, each pixel of the matrix corresponds to a Hounsfield number. The chromatic depth is usually 8 bits (256 gray levels), and the grayscale is similar to the radiographic one, greater attenuation = white; less attenuation = black. Even in CT images, as in radiographic ones, there is a certain amount of "noise." The noise in TC can be distinguished in” quantum noise" And" electronic noise. "
The noise quantum, as for the radiographic examination, is the expression of the probabilistic nature of the interaction of ionizing radiation with matter. Its incidence is, however, lower than in the X-ray examination.
The electronic noise, however, is due to the approximation of the calculation procedures that govern the reconstruction of the images.
It is important to underline that "behind" the chromatic variations of a CT image, there are and remain numbers, numbers that can be read at any moment, and that can give objective quantitative information on the density of the considered fabric. For this reason, a series of operations are possible on the CT images, after data acquisition, which goes under the name of post-processing (variations of the grays represented, linear, angular, density measurements, planar, 3D reconstructions, etc.).
The display mode of the CT images can be varied and adjusted in such a way as to enhance or suppress information present in the images themselves. It is thus possible to discriminate even small density differences (up to 0.5%) and represent them with different gray levels. This results in a higher contrast resolution than the RX exam.
Axial scans and planar reconstructions
The CT exam provides tomographic images. These are usually axial, that is, they cut the median sagittal axis perpendicularly. They, as we said, eliminate the problem of overlap between the various anatomical structures considered. The sequential evaluation of the axial images of a study allows us to have an overall picture of the anatomical district investigated.
The series of axial images can be reconstructed in order to obtain images along different anatomical planes (sagittal, dorsal, oblique, or even curved) (MPR = Multi Planar Reformation).
3D reconstructions
In addition to planar reconstructions, many post-processing software allows obtaining 3D reconstructions. There are several reconstruction protocols. The most used ones are Maximum Intensity Projection (MIP), Volume Rendering (VR), and Surface Rendering (SR). 3D reconstructions for virtual endoscopies and, for the heart, 4D is possible with the most modern CT devices because they include time.
Author: Vicki Lezama