Automated detection of white matter abnormality and its application to Alzheimer's disease

Diffusion tensor imaging (DTI) technique, which can be used to research the white matter of human brain noninvasively, provides more valuable information in the study of white matter abnormality, especially using fractional anisotropy (FA) images calculated from DTI data. Although there are many different DTI data analysis software, most of them are inconvenient to use and qualitative via visual inspection. To provide assistance to physicians in improving the facilitation and sensitivity of the white matter abnormality analysis, we developed an automated FA analysis method of an individual in comparison with a group of normal controls, which was based on the principle of voxel based analysis (VBA), and contained preprocessing, database, and statistical analysis procedures. Only using FA and B
_o
images, the final results about the white matter structure difference would be displayed automatically and interactively. Some data generated during the processing were stored to corresponding path and could be checked when necessary. We applied this methods to Alzheimer's disease (AD). The final result supplied the brain regions where FA values of AD were reduced. DTI data from fifteen AD and sixteen elderly healthy subjects was used to analyse and the brain regions which affected by AD were consonant in the main with the previous study, such as corpus callosum, cingulate regions, and so on. This method could be valuable tool to help physicians in making their clinical decisions.