Sparse-Angle CT Reconstruction Edge-Preserving Algorithm Based On Total Variation

In recent years,sparse angle imaging has been widely used in interventional imaging as well as radiotherapy,and the study of sparse-angle CT is of great significance for reducing radiation dose in clinical treatment.However,the increased noise level and incomplete projection data in sparse-angle CT make serious streak artifacts in the results of sparse-angle CT reconstruction images and degrade the image quality.According to the theory of compressed sensing,a regularization prior can be introduced into the iterative reconstruction algorithm to improve the reconstructed image quality.Therefore,the research objective of this paper is to address the problem that the existing sparse angle CT reconstruction algorithm cannot combine both noise removal and edge-preservation performance,establish a double-constrained sparse prior reconstruction model based on regularization,and propose a corresponding iterative optimization algorithm to solve the sparse-angle CT model to improve the image reconstruction quality of sparse-angle CT.In this paper,we propose a double-constrained iterative reconstruction algorithm for sparse image reconstruction based on the total variation minimization reconstruction model with step effects and excessive image smoothing based on the total variation（TV）algorithm,combined with the edge-preserving performance of the alternating edge-preserving diffusion smoothing algorithm（AEDS）.By using the TV and AEDS algorithms,the reconstructed images are sparsely regularized to better suppress noise and achieve edge-preserving effects.It is demonstrated experimentally that the algorithm can better preserve the image edges and eliminate streak artifacts.In addition,the alternating edge-preserving diffusion smoothing algorithm can effectively protect the image edges due to the high sparsity of l₀ norm.Based on this,we introduce the gradient fidelity term in the alternating edge-preserving diffusion smoothing algorithm to make full use of the information before and after the iteration,and combine the high sparsity of l₀ norm to better preserve the image edge information and eliminate the streak artifacts quality.