A soft-threshold filtering EM-TV algorithm for CT reconstruction

The expectation maximization (EM) algorithm based on statistical property can reduce image noise and artifacts in CT reconstruction. Considering its slow convergence and unsmooth of reconstruction image, this paper proposed EM-TV algorithm which is combined with the minimization of total variation (TV) and a soft-threshold filtering. To make the data smooth and do not loss the information of the edge, first, using EM algorithm reconstruct an image; then its discrete difference transform (DDT) is calculated and the soft-threshold filtering is updated. Finally, experimental results and comparisons with the EM algorithm of Shepp–Logan phantom show that the proposed algorithm can not only make the image of reconstruction smooth and reduce the normalized mean square distance, but also speed up the convergence. And real projection data indicate that this algorithm is effective.