Spectral CT Reconstruction Algorithm Based On Improved Total Variation

Spectral computed tomography(Spectral CT)can distinguish photon energy and obtain projection under multiple energy channels.Since most photon count detectors can only withstand a limited count rate,and a single energy channel contains only a small fraction of the total number of photons,the multi-channel projection usually contains a large noise.So the researchers suggest can effectively restrain various noise spectrum CT algorithm,including the variational as one of the most widely used in CT image reconstruction algorithm,although can well protect the edges of the image,but the reconstructed image will appear in the block artifact or ladder-like pseudo edge,therefore,in order to solve this problem,this paper main research work is as follows:(1)The model of directional total variational(DTV)based on soft threshold filtering is proposed.This algorithm calculates the TV of image from multiple directions,and the DTV is solved by soft threshold filtering,which improves the effectiveness of the TV algorithm.The effectiveness of the algorithm is verified by the simulation experiment of simple model and the actual experiment of clinical mice.(2)A nonlocal TGV reconstruction algorithm based on prior image constrained compressed sensing(PICCS)is proposed.Combining the total generalized global variation(TGV)that can approximate arbitrary order functions,the idea of non-local mean value algorithm and the correlation of reconstructed images under different energy spectrum channels,the high-quality full-energy spectrum reconstructed images were taken as the prior images to guide the CT reconstruction of energy spectrum.Experimental results show that the proposed algorithm can effectively recover image details and edge information while suppressing noise,and the algorithm has good convergence.