Optimization of image preprocessing for diffusion tensor magnetic resonance imaging

Diffusion tensor magnetic resonance imaging (DT-MRI) data should be preprocessed before they are used to analyses for searching white matter fiber structure of human brain. Fractional anisotropy (FA) calculated by DTI data has been widely used because it can describe the anisotropy prosperity of white matter. This paper presents some evaluation criterions to optimize the image preprocessing procedure. Spatial normalization and smoothing are main steps in the procedure. For assessing the normalization, we compute the coherence of principal eignvectors after they are normalized. To test the ideal smoothing parameters, we use Shapiro-Wilk (SW) test to exam the normality of smoothed FA. Based on our assessment, suitable methods and parameters will be achieved and the optimized image processing procedure can help us know more information about white matter changes. In this paper, we used these methods to analyze DTI data of Alzheimer disease patients and matched normal controls.