Research On MRI Algorithm Based On Partial Acquisition And Parallel Imaging

Magnetic resonance imaging(MRI)is an important modern medicine technology for clinical diagnosis.It has many advantages compared with other medical imaging technologies,such as non-radiation,arbitrary cross-sectional imaging,and multi-parameter images.However,its relatively long imaging time limits its development and application in many important medical clinical and research fields.Partial Fourier imaging and parallel imaging can both shorten the scanning time and accelerate the imaging speed of MRI.Partial Fourier imaging methods use K-space conjugation symmetry to acquire more than one half image data to complete image reconstruction quickly,and the scanning speed is nearly doubled,but acceleration factor of this method is limited,maximum acceleration factor is two.The parallel imaging method uses multiple phased array imaging coils to collect image data in parallel at the same time.When acceleration factor of the K-space image under-sampling increases,the image artifacts increase significantly and the image noise is serious.By comparing multiple fast imaging methods such as partial Fourier imaging,parallel imaging,and partial Fourier parallel imaging,this paper mainly studies how to effectively integrate partial Fourier imaging and parallel imaging algorithms to get high quality image.So how to increase MRI speed is still a key problem in MRI research field.First of all,we optimize the POCSENSE method in the parallel imaging method in this paper,and propose a novel PC-POCSENSE algorithm.The reconstruction results show that the PC-POCSENSE method can reconstruct images with less image artifacts and higher SNR under larger acceleration factor.Secondly,in this paper,the phase-constrained PFPI method is proposed by combining the partial Fourier imaging POCS method and the PC-POCSENSE parallel imaging method in sequent order.This method reduces the number of iterations.Finally,by fusing POCS and PC-POCSENSE in the whole iteration process,an iterative PFPI method is proposed to reconstruct partial Fourier parallel under-sampling data.At the beginning of the iteration,the magnitude image of the low-resolution image is input as the initial image,and the dual constraints of phase constraints and region of interest constraints are applied in the method.Moreover,the projection operator of the data convex set is also optimized in this method.More priori information is considered in the reconstruction process,so the number of iterations is greatly reduced in the new algorithm and the reconstruction speed is significantly improved.In the same time SNR of the reconstruction image is larger,and the artifact power(AP)is smaller.Inorder to verify the proposed method,under-sampling brain under-sampling data and knee data is used in the experiment.The reconstruction results show that under the same acceleration factor,the algorithm can reconstruct faster than the traditional partial Fourier and parallel imaging algorithm speed,higher SNR and smaller AP value.This method can shorten more imaging time,and ensure the quality of the reconstructed images.So the purpose of rapid imaging is achieved.