Research On Cone-beam CT Reconstruction And Ring Artifact Correction Algorithms

Computing Tomography (CT) is one of the most important tools for Medical Diagnosis and Non Destructive Testing. Cone-beam CT has been widely used in aviation,aerospace, medical, mechanical and other fields with its advantages of high resolution, low imaging cost and fast imaging. In this paper, FDK reconstruction algorithm and ring artifact correction method are studied.1. In this paper, we first introduce the generation and characteristics of X-ray, and then introduce the basic theory of CT reconstruction, such as Beer’s law, Radon’s positive and inverse transformation and projection theorem. Then, the reconstruction algorithm such as back-projection, filtered back-projection and fan beam equidistant FBP are analyzed and deduced. The principle and general reconstruction steps of the reconstructed algorithm of filtered back-projection are summarized, which lays the foundation for the research of sequencing.2. The FDK reconstruction algorithm is deduced theoretically, and analyzes the process of its discretization. According to the characteristics of the FDK reconstruction process, an improved FDK algorithm which is based on polar coordinates and the periodic characteristics of trigonometric function is proposed. The improved algorithm can reconstruct the projection data of 144 projection data at one tiime, and greatly reduce the calculation of the trigonometric function. Simultaneously, during the transformation of reconstructed image from polar coordinate to Cartesian coordinate, several pixels could be switched once using the symmetry of cotangent function. Experimental results shows that, compared with the traditional FDK reconstruction algorithm, this optimal method could improve the speed by about 10 times.3. In order to eliminate the problem of ring artifacts due to the instability of the detector,the causes of ring artifacts and their characteristics are analyzed. At the same time, one method of multi-threshold based on Otsu method was proposed based on the study of a one-way adaptive smoothing algorithm. Experiments results show that this improved method can both effectively remove the ring artifacts in the case of guaranteeing the details of the image.