弥漫性轴索损伤（DAI）是指头部在快速旋转、直线加速或减速过程中，因产生的剪切力作用于脑白质，造成轴索断裂、降解。该损伤的法医学鉴定难点在于常规轴索损伤免疫组化标记物在DAI与缺血/缺氧性（HI）轴索损伤的鉴别及准确推断生前损伤时间方面存在局限性。本研究拟通过建立DAI和HI轴索损伤大鼠模型并收集人体标本，应用FTIR光谱技术对各脑白质进行化学成分分析,为阐释两种轴索损伤的病理生理过程提供实验依据；采用遗传算法、主成分分析和偏最小二乘法筛选特征性和时序性光谱变量，建立BP人工神经网络模型，用于客观推断DAI生前损伤时间及与HI轴索损伤的鉴别；此外，还收集1个月内死后各时间点大鼠脑白质光谱数据，建立多元线性回归方程，校正死后自溶对光谱分析的影响。本课题探索了生物光谱学在法医学DAI诊断中应用的可能性，为突破常规法医病理学诊断技术难点提供了新的思路。

Diffuse axonal injury (DAI) is characterized by axonal disruption and degradation, mainly resulting from mechanical shearing forces on the white matter during rapid rotation, acceleration/deceleration of the head. Immunohistochemical methods, commonly used methods in forensic practice, is limited in differentiating DAI from hypoxic-ischemic (HI) axonal injury and estimating the survival interval of DAI. Therefore, the aim of this project is to utilize fourier transform infrared (FTIR) spectroscopy to detect biochemical changes in the white matter of animal and human samples with DAI and HI axonal injury. This study would provide insight into understanding of pathophysiology of both types of axonal injury. In order to estimate DAI intervals and distinguish DAI from HI axonal injury, error back-propagation artificial neural network (BP-ANN) models are established based on time-dependent and specific spectral variables, which are determined by variable selection methods including genetic algorithm, principal component analysis and partial least square-discriminant analysis; a multiple linear regression equation is calibrated by spectral data available from the white matter of rats within a month postmortem, and used to minimize the influence of autolysis on spectral analysis. The study would demonstrate the possibility of biospectroscopy as a powerful tool to diagnose DAI and estimate survival interval of DAI, and pave a way for accurate DAI diagnosis in forensic pathology.