Research On Regularization Method For Sparse-view CT Reconstruction

The key factors which influence the accurarcy of CT reconstruction is acquistion of projection data and method of reconstruction.Affected by the actual object detection environment and structure of objects,the detector can only get a small amount of uniformly-spaced projection data,which is called sparse-view CT.Thus,it is of great value to study the reconstruction algorithm suitable for sparse-view CT.It is an effective method to improve the quality of the reconstructed image by adding regularization term in the iterative reconstruction algorithm,which take the sparsity of image as a priori knowledge.According to the regularization theory,this paper researchs a new sparse-view CT reconstruction model based on spatiotemporal tensor framelet and low-rank matrix recovery technique.To solve the mixed norm nondifferentiable problem,use the Split-Bregman algotithm “split” reconstruction model into several differentiable subproblems.Compared with the ART-TV algorithm,the experimental results show that the proposed algorithm is feasible to improve the accuracy and noise immunity.X-ray spectrum is continuous with different energy segments which is produced by X-ray source,but reconstruction algorithm ignore the spectrum of X-ray,which is a mismatch problem.This paper researchs the two mian factors that affect the X-ray attenuation : the Compton scatter and photoelectric,and the attenuation coefficient is expressed as a linear combination of them.At the same,the principle of multi spectral CT imaging is studied.Considering the Compton scatter componnt and the photoelectric component are indepent of the energy of X-ray,this paper introduce a spectral inerior CT projection decomposition algorithm based on spatiotemporal tensor framelet: assuming photoelectric component and Compton scatter component are piecewise smooth.The photoelectric component and Compton scatter componentare separated from the multi spectral projection to reconstruct the image of the object under the reference energy.The simulation results show that this method can effectively reduce the image beam-hardening and improve the contrast of the reconstructed images.