Research On The Low-dose CT Reconstruction Algorithm Based On ADMM

CT reconstruction at sparse angle is one of the effective methods to reduce the dose of X-ray radiation.With the development of compression perception theory,iterative reconstruction algorithm plays an important role in sparse angle CT reconstruction,which breaks the traditional bottom line of sampling law and can be used in sparse sampling.The reconstruction quality of the image is higher than that of the analytical algorithm.However,according to the basic model of compression perception theory,different regular terms will lead to different reconstruction accuracy.In order to further reduce the radiation dose of X-ray,we want to study how to reconstruct better image quality with fewer projection angles.Aiming at the problem of CT reconstruction from sparse angle,this paper analyses the basic model of CT reconstruction based on compressed sensing theory,and studies two different regularization terms of CT reconstruction model.Studied the sparse angle CT reconstruction model based on the LP regular term,and the model is solved by using the ADMM algorithm and the GST algorithm.Simulation and actual data,projection data with noise and without noise are used to verify and analyze the algorithm,and the visual and numerical comparison between we studied algorithm and ART-TV,ART-LP,Split-Bregeman-LP algorithm is compared.The experimental results show that the images obtained by the ADMM-LP algorithm are clearer and more complete.Then,in ADMM-LP algorithm,because of the calculating that the product between a large sparse matrix and a large sparse matrix,the computing cost of the algorithm is large.An accelerated algorithm is studied.The algorithm has no obvious precision reduction but it effectively reduces the time consumed.At last,Combined with the Mumford-Shah functional in thefield of image restoration,a mathematical model based on TGV canonical term and Mumford-Shah functional for CT reconstruction is studied.Then the model is solved effectively by ADMM algorithm.Finally,the model is solved effectively by ADMM algorithm.Simulation data are used to verify the algorithm,the results show that the algorithm studied in this chapter has certain advantages in edge protection.