A new fast and parallel MRI framework based on contourlet and compressed sensing sensitivity encoding (CS-SENSE)

摘要

Compressed sensing MRI (CS-MRI) and compressed sensing sensitivity encoding (CS-SENSE) only include two regularization items, total variation (TV) and Wavelet, which leads to artifacts remaindering in 1-D random sampling. In order to improve the performance of them, a new regularization item-Contourlet is introduced to constrain the solution with the other two regularization items together. The solution of the new model is searched by a modified fast composite splitting algorithm (FCSA), named mFCSA. Experimental results show the new framework and its mFCSA implementation have good performance in artifacts reducing.