摘要
基于最大似然期望法(ML-EM)重建的图像质量好,但因其收敛速度太慢,而难以直接应用于临床.有序子集最大期望法(OSEM)具有较高的重建图像质量和较短的计算时间,超松弛因子(OR)有可灵活调节超松弛因子和加速收敛两种功效.文中将OSEM和OR有机地结合起来用于PET图像重建,形成可变超松弛因子的OSEM重建算法(OR-OSEM).该方法综合了上述两种算法的优点,具有可根据需要灵活改变子集数和超松弛因子的功能.计算机仿真模拟和临床PET投影数据的重建结果表明了该方法的有效性.
Although ML-EM is a perfect reconstruction algorithm, its slow convergence rate obstructs its direct use in practice. OSEM has good reconstructed image quality and uses less computation time. Overrelaxation technique has adjustable relaxed parameters and speeded convergence function. A variable overrelaxation OSEM （OR -OSEM） algorithm with both advantages is proposed in this paper. It can change order set numbers and overrelaxation parameters expediently according to real need. Resuits of computer simulated data and clinical PET projection data demonstrated that this algorithm could be used to get good quality image.
出处
《武汉理工大学学报（交通科学与工程版）》
2007年第1期111-114,共4页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
国家973计划项目资助(批准号:2003CB716102)
