Department of Neurology, University of Wisconsin, UW Medical Foundation Centennial Building, Madison, WI 53705, USA.
J Neurosci Methods. 2012 Apr 30;206(1):40-5. doi: 10.1016/j.jneumeth.2012.02.004. Epub 2012 Feb 11.
A statistical approach is presented for the quantitative analysis of diffusion tensor imaging (DTI) directional information using Fisher statistics, which were originally developed for the analysis of vectors in the field of paleomagnetism. In this framework, descriptive and inferential statistics have been formulated based on the Fisher probability density function, a spherical analogue of the normal distribution. The Fisher approach was evaluated for investigation of rat brain DTI maps to characterize tissue orientation in the corpus callosum, fornix, and hilus of the dorsal hippocampal dentate gyrus, and to compare directional properties in these regions following status epilepticus (SE) or traumatic brain injury (TBI) with values in healthy brains. Direction vectors were determined for each region of interest (ROI) for each brain sample and Fisher statistics were applied to calculate the mean direction vector and variance parameters in the corpus callosum, fornix, and dentate gyrus of normal rats and rats that experienced TBI or SE. Hypothesis testing was performed by calculation of Watson's F-statistic and associated p-value giving the likelihood that grouped observations were from the same directional distribution. In the fornix and midline corpus callosum, no directional differences were detected between groups, however in the hilus, significant (p<0.0005) differences were found that robustly confirmed observations that were suggested by visual inspection of directionally encoded color DTI maps. The Fisher approach is a potentially useful analysis tool that may extend the current capabilities of DTI investigation by providing a means of statistical comparison of tissue structural orientation.
本文提出了一种基于 Fisher 统计的方法，用于对扩散张量成像（DTI）的方向信息进行定量分析。Fisher 统计最初是为古地磁学领域的向量分析而开发的。在这个框架中，基于 Fisher 概率密度函数（正态分布的球形模拟），已经制定了描述性和推断性统计。Fisher 方法用于研究大鼠脑 DTI 图谱，以表征胼胝体、穹窿和背侧海马齿状回门区的组织方向，并比较癫痫发作（SE）或创伤性脑损伤（TBI）后这些区域与正常大脑的方向特性。为每个脑样本的每个感兴趣区域（ROI）确定方向向量，并应用 Fisher 统计来计算正常大鼠和经历 TBI 或 SE 的大鼠的胼胝体、穹窿和齿状回的平均方向向量和方差参数。通过计算 Watson's F 统计量和相关的 p 值来进行假设检验，给出了分组观察结果是否来自相同方向分布的可能性。在穹窿和中线胼胝体中，各组之间没有检测到方向差异，但是在门区，发现了显著差异（p<0.0005），这有力地证实了通过对方向编码彩色 DTI 图谱进行目视检查所提示的观察结果。Fisher 方法是一种潜在有用的分析工具，它可以通过提供组织结构方向的统计比较手段，扩展 DTI 研究的现有能力。
