FALSE DISCOVERY RATE ANALYSIS OF BRAIN DIFFUSION DIRECTION MAPS.

Area of Science:

• Neuroimaging
• Biostatistics
• Medical Physics

Background:

• Diffusion Tensor Imaging (DTI) noninvasively maps brain white matter by measuring water diffusion directions, which serve as proxies for neural fiber orientation.
• Analyzing differences in white matter structure between subject groups is crucial for understanding neurological conditions.

Purpose of the Study:

• To develop and validate statistical inference methods for detecting significant differences in diffusion directions between two subject groups using DTI data.
• To improve the accuracy and power of statistical tests in large-scale neuroimaging data analysis.

Main Methods:

• Utilized a Watson model for directional data to compute a test statistic for diffusion direction differences at each brain location.
• Employed false discovery rate control for selecting significant regions.
• Incorporated empirical null density estimation and local spatial averaging to enhance statistical power and refine null distribution modeling.

Main Results:

• The proposed methods successfully identified brain regions with differing diffusion directions between the studied groups.
• Empirical null density and spatial averaging led to substantial improvements in statistical power.

Conclusions:

• The developed statistical inference framework offers a robust approach for analyzing DTI data to detect group differences in white matter architecture.
• These methods are broadly applicable to other large-scale simultaneous hypothesis testing problems with spatial structures.