The Study Of Key Points Of Parallel MR Imaging By Using Sensitivity Encoding (SEMSE) Method

Parallel MR imaging (pMRI) has been one of the most successful and commerciallyavailable approaches to reduce acquisition times, which uses an array of receiver coils toacquire multiple sets of under-sampled k-space data simultaneously. GRAPPA and SENSE aretwo typical different reconstruction methods. GRAPPA is one kind of algorithm which is basedon K Space Field, and SENSE is the other one based on Image Field. In this paper, one newapproach based nonlinear GRAPPAand SC-SENSE is proposed to implement the algorithms ofpMRI, using different kinds of sampling methods. In addition, another SENSE algorithmapproach based on Compressed Sensing and FCSAis also proposed. The main researches are:(1)In this paper, a new hybrid reconstruction method is proposed to implement the pMRIalgorithms, based on combining algorithms of GRAPPA and SENSE. In this method, thenonlinear GRAPPA method is applied to calculate the sensitivity map of each coil firstly, andthen SENSE method is adopted to reconstruct the MR image from the under-sampled K-spacedata. The proposed hybrid technique is tested on MR brain image reconstructions at variousacceleration rates. The experiment results show that the proposed method is capable of achievingsignificant improvements in reconstruction accuracy when compared with the GRAPPA andSENSE methods, with lower artificial power (AP) and higher signal to noise ratio (SNR).(2)Another new parallel magnetic resonance imaging algorithm based on sparse constrain isalso proposed to implement the pMRI algorithms in this paper. This algorithm can reconstructimage from sub-sampled K space data. In sparse-enforced sensitivity encoding (SENSE)reconstruction, the optimization problem involves a number of L1-norm regularization terms (i.e.total variation or TV, and L1norm). Due to the non-smooth nature of the regularization terms,the optimization problem is difficult to solve while reconstructing the MR image. In this paper,in order to effectively solve the optimization problem of the Sparsity-regularized SENSEreconstruction, a new method, called the FCSA (fast composite splitting algorithm, FCSA) was proposed in this paper. The FCSA algorithm decouples the large optimization problem into TVand L1sub-problems, and the sub-problems can be solved by existed methods. At last, based onSENSE framework, the MR image can be obtained. The FCSA-based parallel MRI technique istested on MR brain image reconstructions at various Acceleration rates and with differentsampling trajectories. The results indicate that, for Sparsity-regularized SENSE reconstruction,the FCSA-based method is capable of achieving significant improvements in reconstructionaccuracy when compared with the typical NLCG reconstruction method.