一种用于估算磁共振成像中水分子扩散系数的非迭代算法(英文)  

A Non-iterative Algorithm to Estimate Water Molecule Diffusion Coefficient in Magnetic Resonance Imaging

作　　者：高嵩[1] 王飞[1] 

机构地区：[1]北京大学医学部医用理学系

出　　处：《航天医学与医学工程》2012年第4期296-298,共3页Space Medicine & Medical Engineering

基　　金：Supported by National Natural Science Foundation of China(81171130);Natural Science Foundation of Beijing(7102102)

摘　　要：目的提出一种更加可靠和高效的算法,用于计算磁共振扩散加权成像(DWI)中人体内水分子的扩散系数。方法根据扩散加权成像中磁共振信号的双指数衰减规律,我们提出了双线性拟合算法以替代现在广泛使用的Levenberg-Marquardt迭代算法。使用两次线性拟合分别计算快慢扩散系数,整个计算过程中不必估算初始值。结果双线性拟合算法计算所用时间远少于Levenberg-Marquardt算法,计算结果合理。而Levenberg-Marquardt算法的结果中有大量额外解。结论与Levenberg-Marquardt算法相比,双线性拟合算法在分析磁共振扩散加权成像数据方面不仅算法可靠而且计算效率高。Objective To propose a more reliable and time-efficient algorithm to estimate the diffusion coeffi- cients of water molecular in magnetic resonance diffusion weighted imaging （ DWI ） experiments in vivo. Methods According to the novel feature from the hi-exponential attenuation of diffusion weighted magnetic res- onance signal, the authors proposed the algorithm of twice-linear-fitting instead of the commonly used iterative Levenberg-Marquardt method. This algorithm consisted of two liner fitting steps to estimate the fast and slow diffusion coefficients respectively. It was unnecessary to estimate the initial value in the whole fitting process. Results The time consumption of the twice-linear-fitting algorithm was much less than that of the Levenberg- Marquardt method. All results from the examples were reasonable. However, there were extraneous solutions in the results of Levenberg-Marquardt algorithm. Conclusion Compared with the Levenberg-Marquardt meth- od, the twice-linear-fitting algorithm is a reliable and more time-efficient approach to estimate the diffusion co- efficient in magnetic resonance diffusion weighted imaging.

关 键 词：线性拟合 扩散加权成像 非迭代 

分 类 号：R445.2[医药卫生—影像医学与核医学]