中文摘要：

通过核磁共振扩散张量成像（DTI）得到的特定值域的扩散各向异性指数（DAI）可用于揭示水分子扩散椭球的形态学特征，定量反映被成像物体内部水分子扩散的优势方向和强度，间接得到被成像物体内部的组织结构信息. DAI的可靠性直接影响对DTI数据的分析和理解.本文基于扩散张量椭球的几何学信息，提出利用扩散椭球几何比（EGR）定量描述水分子扩散的各向异性程度.通过蒙特卡罗模拟实验和对人脑DTI数据进行分析，并与当前广泛应用的水分子扩散各向异性分数（FA）和近期文献提出的扩散椭球面积比（EAR）进行对比.实验发现EGR在不同级别噪声影响下的对比度效果和抗噪性都优于FA及EAR.而且EGR加入了体积修正，增强了盘形扩散张量情况下的敏感性，能够更好地鉴别神经纤维束交叉情况，对于各向异性扩散程度较高的白质深层和相对均质的表层都有较好的量化区分结果.

英文摘要：

Diffusion anisotropy indices （DAIs） are parameters derived from diffusion tensor imaging （DTI） data which describe the mor-phological characteristics of diffusion tensor within a specific range. DAIs are the measurements used to quantitatively describe the diffusion direction and strength of the hydrone in vivo, so that DAIs enable one to indirectly probe the internal structure of an imaging subject. The reliability of DAIs is of great importance for the analysis and interpretation of DTI data. Based on the geometric char-acteristic of the diffusion tensor ellipsoid, we propose a new DAI, the“ellipsoidal geometric ratio”（EGR）, to describe the hydrone diffusion anisotropy property. The analysis results of Monte Carlo simulation and human brain DTI data show that the EGR has better contrast and robustness than fractional anisotropy, the most commonly used DAI, and ellipsoidal area ratio at different noise levels. Furthermore, since EGR makes full use of the ellipsoidal volume information, it is more robust than any other DAIs in the fiber crossing case. EGR may be a superior measure of diffusion anisotropy both in quantifying deep white matter with relatively high anisotropy and pericortical white matter with relatively low anisotropy.