Evaluating the accuracy of diffusion MRI models in white matter

Area of Science:

• Neuroimaging
• Biophysics
• Computational Neuroscience

Background:

• Diffusion MRI models are crucial for inferring tissue properties and white matter tract orientation.
• Accurate model fitting is essential for reliable diffusion MRI data analysis.
• Previous evaluations of commonly used diffusion MRI model accuracy were lacking.

Purpose of the Study:

• To evaluate and compare the model accuracy of the diffusion tensor model (DTM) and sparse fascicle models (SFM).
• To assess model performance across different b-values in white matter regions.
• To determine which model provides more accurate fiber orientation distribution function (fODF) estimation.

Main Methods:

• Cross-validation was employed to assess model accuracy.
• Models were fitted to diffusion MRI data from white matter voxels.
• Model predictions were validated against independent datasets at varying b-values.

Main Results:

• SFM demonstrated higher model accuracy than DTM across most white matter regions.
• SFM outperformed DTM, particularly at b-values > 1000, in areas with fiber crossings and around optic radiations.
• SFM exhibited superior parameter validity, yielding more accurate fODF estimations.

Conclusions:

• SFM provides a more accurate representation of diffusion MRI data compared to DTM.
• SFM's enhanced accuracy and parameter validity are particularly beneficial for analyzing complex white matter architecture and improving tractography.
• This study provides the first comprehensive evaluation of DTM and SFM model accuracy in diffusion MRI.