FRATS: Functional Regression Analysis of DTI Tract Statistics

Area of Science:

• Neuroimaging
• Biostatistics
• Developmental Neuroscience

Background:

• Diffusion Tensor Imaging (DTI) offers in vivo insights into white matter microstructure and tissue properties.
• Analyzing multiple diffusion properties along fiber bundles requires advanced statistical methods.

Purpose of the Study:

• To present FRATS, a functional regression framework for analyzing diffusion properties along fiber bundles.
• To associate these diffusion properties with covariates like age, diagnostic status, and gender.
• To characterize neurodevelopmental changes in white matter tracts.

Main Methods:

• FRATS integrates local polynomial kernel smoothing, a functional linear model, a global test statistic, and resampling for p-value approximation.
• The method analyzes five diffusion properties (fractional anisotropy, mean diffusivity, etc.) along specific white matter tracts.
• Applied to a clinical study of neurodevelopment in the splenium of the corpus callosum and right internal capsule.

Main Results:

• Significant effects of age and gestational age on five diffusion properties were identified in both analyzed tracts.
• The FRATS framework successfully characterized developmental trajectories of diffusion properties.
• Demonstrated the association between diffusion properties and developmental factors.

Conclusions:

• The FRATS pipeline provides a robust method for analyzing complex diffusion MRI data.
• It enhances understanding of normal brain development and the neural basis of neuropsychiatric disorders.
• Facilitates research on environmental and genetic influences on white matter development.