相位解缠绕算法在磁敏感加权图像中的应用与比较

摘要

为了找到能有效解决磁敏感加权成像(SWI)相位信息缠绕问题的方法,针对8种空域解缠算法进行分析和比较,其中:选取5种全局优化算法,分别为加权最小二乘法(WLS)、解缠离散估计的相位区域扩张标记法(PRELUDE)、最大流/最小切相位解缠法(PUMA)、差值映射累加法(ARM)、排序多聚类法(SL-MC);3种积分算法,分别为枝切法(Branch-cut)、基于傅里叶滤波的质量图引导法(WFF-QG)、递归正交参考相位解缠法(PUROR).采用两组Matlab仿真图像和两组来自Alltech公司提供的1.5T真实SWI数据进行实验,以原始缠绕相位与再缠绕相位之间的突变点统计值和算法运行时间作为量化指标,评价8种算法的优越性.结果显示,PUROR的P±|M|值为0,运行时间最短,鲁棒性较强,解缠绕结果可靠,在实际临床应用中可尝试采用PUROR法.%In order to develop an effective method to solve the phase unwrapping problem in the susceptibility weighted imaging (SWI),this paper analyzed and compared eight kinds of spatial domain unwrapping algorithms,among which five global optimization algorithms were selected,including WLS (weighted least square),PRELUDE (phase region expanding labeler for unwrapping discrete estimations),PUMA (phase unwrapping max-flow/min-cut),ARM (accumulation of residual maps),and SL-MC (sorted list,multiclustering);and three integral algorithms were selected,including Branch-cut,WFF-QG (windowed Fourierfiltered and quality-guided method),and PUROR (phase unwrapping using recursive orthogonal referring).Experiments were carried out by using two sets of MATLAB simulation images and two sets of 1.5T real SWI data provided by Alltech.The reliability of the unwrapping methods was evaluated by using differences between wrapped and re-wrapped phase and time.The results show that the P ± | M | value of PUROR was zero and it had the fastest running time,which implied the application potential in clinical practices.