Research On The Application Of Split-Bregman Algorithm In CT Reconstruction

Based on Compressed Sensing theory and Discrete Gradient Transform, it has been proved that both the interior problem of computed tomography and Sparse-view problem can be expressed as a total variation(TV) minimization problem. At the same time, the SplitBregman algorithm shows great advantages in solving TV minimization problems. The paper studies the interior problem of CT and the problem of reconstruction for sparse-view CT based on Split-Bregman algorithm. The main contents are as follows:(1) Derives the Split-Bregman algorithm for CT reconstruction, and studies the selection of the regularization parameters for Split-Bregman algorithm. Then, compares the SplitBregman algorithm with the traditional TV minimization methods, Gradient Descent method and Soft-Thresholding algorithm, and the experimental results show that Split-Bregman algorithm shows significant superiority in both the convergence speed and the quality of reconstructed image comparing with traditional methods.(2) For the calculation-consuming problem that Split-Bregman algorithm needs to perform several forward/ back-projection in each iteration, the paper uses the ordered-subset(OS)method to accelerate it and gives an OS-Split-Bregman(OS-SB) method. Then applies the OS-SB algorithm to the TV minimization model of the interior problem of CT and Sparse-view problem, respectively, both numerical simulations and clinical applications are performed and the results confirm the validity of the OS-SB method.