Research And Development Of X-Ray Spectral CT And Image Reconstruction Algorithm

In recent years,with the continuous improvement of computer hardware performance and algorithm innovation,industrial X-ray CT image reconstruction technology has been rapidly developed.X-ray CT(XCT)uses the attenuation information carried by X-rays after penetrating the material to reconstruct the tomographic image reflecting the internal structure of the object under test through algorithms,which brings rapid development to the study of the internal structure of the object and the detection of material defects.However,the traditional XCT is faced with the problems of poor tissue/material contrast,low resolution,inability to perform quantitative analysis,low material identification,etc.The X-ray spectral CT technology is thus born,and the X-ray spectral CT technology acquires the projection data of the object in several different energy channels by photon counting detector at one time,which has the advantages of high resolution,high material identification,and quantitative analysis of the tissue/material.This paper further optimizes the design of the industrial CT system structure based on the previous research,including the use of Geant4 to model the system environment of ray source,1*128 pixel line array CZT detector in the first stage,and the complete process of CT data acquisition.With the direction of the simulation results,the optimal energy of X-ray is selected and the object is designed.With a CZT photon-counting detection system,the experimental data are collected and then CT images are reconstructed with FBP algorithm.Through the advantage of energy spectrum energy threshold division,better energy region reconstruction results are obtained and combined with the image quality evaluation system to provide a better direction for the optimization of CT system parameter settings.In terms of image artifact removal,the median filtering plus watershed algorithm based on the reconstructed image are proposed,and for the linear artifact removal from projection data,the method of straight line detection and filtering processing based on the Hough transform is proposed,and through experimental verification,these two methods have better artifact removal capability.