Separation of prosthesis projections and tissue implants in spiral CT scan sinogram images using active contour methods

Master's Thesis in Bioelectrical Engineering

Abstract

Separation of prosthesis projections and tissue implants in spiral CT scan sinogram images using active contour methods

Effortlessly

Metal implants such as prostheses and tooth filling materials during the reconstruction of CT images with different methods, cause artifacts in the form of light and dark radial lines. It appears around the metal object and seriously limits the medical value of CT images, as it affects diagnostic and treatment decisions. Projection data that are affected by this artifact are called missing projections and appear in the sinogram image as distinct areas with very high intensity and values. Therefore, if these areas are identified as much as possible, their values ??are replaced with appropriate values, and another image is reconstructed from the improved sinogram, then a high quality image is created. In this dissertation, a method for separating the mentioned areas is introduced. In this way, first, several active contour methods are introduced and the best of them is selected for the approximate separation of the border of the respective areas. Then, for more accurate separation, a generalized Hough transform is defined to find sinusoidal curves. By applying the Hough transformation to the separated border and cross-border points, sinusoidal curves corresponding to these points, that is; The relevant areas are identified. Finally, with an algorithm after processing, this separation ends well. The results of the image reconstruction from the improved sinogram show that the proposed algorithm has been able to identify the areas related to metal objects in the sinogram well.

In this chapter, first, a brief explanation is given about the need for body imaging systems and the emergence of the CT scan machine, then the general process of image creation and the causes of artifacts and distortions caused by metal texture in the image are described, and finally, the general objectives and parts of the thesis are presented.

1-1- Introduction

Maintaining and promoting human health as a sustainable development axis has long been considered by scientists in various fields of science. Acquiring knowledge and gaining skills in the field of using different tools to serve mankind, especially for early diagnosis of diseases and their timely treatment, is and is the daily concern of medical scientists. The production and progress of science and technology has not only provided huge developments in this direction, but has significantly increased the reasons for human desire to use technology.

Getting information about how the body's systems are established and function accurately in normal conditions and their changes due to illness is necessary to understand and justify the occurrence of symptoms and signs. Although anatomy or the knowledge of internal organs and systems through direct contact and cutting of the human body provides comprehensive information to doctors and medical scholars, but relying on the exclusive use of this method to know the physical and functional condition of internal tissues is not always possible, but can cause considerable harm to humans. Using previous experiences on corpses and generalizing individual findings is also not possible due to the considerable differences in the anatomy of the human body. Therefore, the use of tools and technology to obtain accurate information from the inside of humans has been one of the never-ending ideals of medical scholars. Throughout history, the idealism of medical scholars in obtaining more complete information from inside the human body has led them to create, develop and use technology.

Konrad Rongten[1], a German scientist, discovered the existence of X-rays for the first time in 1895 and accidentally discovered the unique capabilities of this ray in imaging the internal organs of the body, especially bones. In this way, a new beginning was made for a great development in the field of medical imaging.

X-ray is a series of electromagnetic rays that covers a wavelength from a few picometers to a few nanometers and passes through the materials with different attenuation coefficients depending on the type of material. This is the same feature that is used in normal radiology. In such a way that x-rays pass through bones and hard tissues and are absorbed more than other tissues, and when they are converted into images by x-ray sensitive films, this attenuation of the rays is seen as bright areas and can be recognized.This process can be seen in figure "1-1".

Figure 1-1: Normal radiology. a) How to place components. b) An example of a chest image [1].

Although conventional radiology has helped medicine a lot and even today it is used in many cases, but the following weaknesses made its transformation and progress inevitable:

High and uncontrolled dose of X-rays, which is harmful to the body.

Compatibility of the three-dimensional volume of the body on a two-dimensional image causes organs to overlap in the image and lower the accuracy of the doctor's diagnosis.

The low contrast of the image is due to the diffusion of the source of radiation and the lack of production of completely parallel rays.

And for this reason, the need for computer radiography devices with city scans [2] emerged [1]. The details of CT scan are described in chapter "2".

1-2- Filtered Back Projection (FBP) and artifacts in CT

After the invention of CT, image reconstruction with FBP method was introduced for physical applications. To date, FBP remains the most widely used image reconstruction method in CT. This method is called a so-called transformation method because it is based on the assumption that the measurements, the Radon transformation of the linear attenuation coefficient distribution and the analytical inverse of the Radon transformation are a direct solution for image reconstruction. If this algorithm to ideal projections; That is, to apply an unlimited number of measurements with unlimited narrow beams and without noise, then this solution is ideal. But in practice it is not like this. Because first, the measurements are read with a limited number of detectors in a limited rotation range and using limited wide beams. Secondly, the x-ray tube emits a continuous spectrum that is created as a result of the hardening of the rays [3]. Thirdly, the measurements are noisy and include scattered radiation. There are many other differences between the actual measurements and the Radon transform, all of which cause artifacts in the reconstructed images. But usually the errors caused by these approximations are relatively small and the FBP results are satisfactory. In addition, artifacts can be reduced by applying corrections before or after applying FBP. For example, noise and artifacts caused by aliasing are reduced by applying a low-pass filter to the raw data. However, under many circumstances, including the presence of high-density objects, the artifacts become incredibly intense. Figures "1-2" and "1-3" show an example of striatal artifacts caused by the presence of metallic objects.

Figure 2-1: CT scan of a patient with a hip prosthesis. (a) Patient with prosthesis. (b) CT image of a cross-sectional surface including a prosthesis that has a striped artifact caused by the prosthesis [3]. Many researchers have tried with different methods to eliminate metal artifacts, and they are usually based on the assumption that the measured information that is affected by the metal objects is either ignored or replaced by interpolation [4] between neighboring measurements. These existing algorithms lead to a drastic reduction of metal artifacts, but some of them create new artifacts. Therefore, better algorithms are needed to reduce metallic artifacts more effectively. In CT topics, the most important causes of metallic artifacts that appear as dark and light lines [5] are: beam hardening, lack of photons [6], scattering and relative volume effect, which are mentioned in the following definitions [2-3].  

Figure 1-3: CT images of the skull, which includes tooth filling materials [3].

1-2-1- Hardening of the rays

In the image reconstruction process, it is assumed that monochromatic X-rays [7] are used. But in practice, the X-rays that are used have a wide energy spectrum [8]. Therefore, when a group of rays is irradiated to objects, those rays that have a higher attenuation coefficient; That is, they have a low energy spectrum and are removed from the group of rays sooner. Therefore, the average energy of the beam increases and its penetration ability increases. Generally, those rays whose energy spectrum is in the limits that are more easily attenuated are called soft X-rays and those that penetrate more into objects are called hard X-rays.