Research On CT Image Ring Artifact Reduction Based On Projection Sinogram

Computed tomography(CT)is an advanced detection technique in modern medical diagnosis and radiation therapy.Human internal tissue structure and lesion status can be accurately and nondestructively obtained by using CT.The density distribution and composition of each tissue can also be quantitatively calculated.Thus,CT is widely used so far.The third generation of scanning CT system is widely used in medical diagnosis and industrial field because of its fast scanning speed.As one of the important problems of the third generation CT scanning system,ring artifacts not only seriously affect the image quality,but also disturb the subsequent image segmentation and noise processing.The diagnosis of images is influenced.Therefore,removal of ring artifacts is a crucial step in CT image processing.CT image reconstruction algorithms are mainly divided into two categories.The first class is analytic reconstruction algorithm.A classic analytic reconstruction algorithm is Filtered Back Projection(FBP).The main steps of this algorithm are as follows:First,the examined object is scanned.Then the data detected by detector is processed by using filter.Finally,the filtered data is backprojected to obtain a CT image.Because of its fast reconstruction,FBP is widely used in CT image reconstruction.The second type of method is the iterative reconstruction algorithm.Among them,Algebraic Reconstruction Technique-Total Variation(ART-TV)proposed by Sidky et al.is one of the most typical iterative reconstruction algorithms.ART-TV algorithm contains two main steps:algebraic iterative reconstruction and total variation minimization.Algebraic Reconstruction Technique(ART)is based on the discrete model.The idea is to discretize the image to be reconstructed first and give an initial image.Then,the projection matrix is calculated based on the reconstruction model and the initial value.Finally,the real projection data measured by the detector is compared to correct the current estimation value.The above process is repeated to gradually approaching the real image.The ART-TV algorithm is based on the ART reconstruction algorithm.Algorithm is improved by being combined with the total variation,which can effectively improve the image quality.In this paper,we first systematically review CT system composition,imaging theory and classical reconstruction algorithms.Then,aiming at the problem that the continuous large area dead bins of the detector leads to the missing of the projection sinogram data,the following two reconstruction algorithms are proposed respectively:Firstly,we propose a practical approach based on sinogram correction to suppress CT ring artifacts on the condition that some elements in a detector are damaged continuously in a relatively large range.In the method,line artifacts with wide width in projection space are corrected using linear interpolation to produce CT image as precorrected image.Next a prior image is generated based on the precorrected image.Then the prior image is used to estimate the missing data of original projection.Finally,CT image is reconstructed by using the corrected projections.Simulation and real data experiment results show that the proposed algorithm significantly reduces ring artifacts and improves the image quality significantly,compared with the traditional linear interpolation method.Secondly,two ART-TV reconstruction methods based on few-view angle are proposed to effectively solve large area data missing in the central of projection sinogram.The first method is the ART-TV reconstruction method based on projection pre-correction.At the first iteration,the dead bins are corrected by using linear interpolation,and sinogram completion for the following iterations before ART-TV algorithm is used.Then,the final CT image is reconstructed by ART-TV algorithm.The experimental results show that this method can produce better quality images than the conventional ART-TV algorithm when the data in the central region of the projection sinogram is missing.However,ART-TV reconstruction based on projection pre-correction cannot remove the artifacts completely when the missing data in the center of the projected sinogram is larger.Therefore,a second method of detector offset is proposed.The detector offset method first moves the detector on the left so that the abnormal region of detector is not symmetric about the center.Lastly,the projection data is reconstructed by using conventional ART-TV algorithm without considering the missing data information.Experiments show that the two proposed methods can effectively solve the problem of missing data in the central area of the projection sinogram.