并行MRI图像重建算法比较及软件实现

摘要

首先介绍了不加速的并行MRI图像重建方法,然后对加速的并行MRI的4种图像重建算法进行了比较,得出结论:加速因子相同时,重建质量上,GRAPPA和SENSE的重建质量最好,SMASH的重建质量次之,PILS算法对线圈位置要求极高,重建质量最差;重建速度上,SMASH的重建速度最快,其次是SENSE和PILS,GRAPPA的重建速度最慢.当加速因子变大时,所有算法重建质量都变差.最后介绍了算法实现软件,该软件可以读入原始数据,显示数据采集轨迹,计箅线圈灵敏度,选择图像重建方法,分析和比较重建图像质量.该软件为我国在MRI成像领域提供了一个学习和进一步研究图像重建算法的有力工具.%This paper reviews the commonly-used image reconstruction algorithms in parallel magnetic resonance imaging. First, the reconstruction algorithms used in paral lel MRI without acceleration was discussed. Then, the commonly-used reconstruction algorithms in parallel MRI with acceleration, SENSE, PILS, SMASH and GRAPPA,were compared. It was shown that, with the same accelerating factor, the quality of GRAPPA and SENSE reconstructions are the best among the four, while that of PILS reconstruction is the worst. In terms of reconstruction speed, SMASH is the fastest,and GRAPPA is the slowest. The performance of all four reconstruction algorithms degraded with increasing accelerating factor, suggesting that increase of imaging speed is at the expense of the cost of SNR. We also implemented the reconstruction algorithms for parallel imaging on an Matlab GUI platform. The resulting software has the functions of reading raw data, showing sampling trajectory, calculating coil sensitivity,choosing image reconstruction method, analyzing and comparing image quality etc. We believe the software will be helpful for study and research on parallel MRI reconstruction algorithms.