SR-HARDI: Spatially Regularizing High Angular Resolution Diffusion Imaging.

Area of Science:

• Neuroimaging
• Diffusion MRI
• Computational Neuroscience

Background:

• High angular resolution diffusion imaging (HARDI) is crucial for mapping brain connectivity and detecting changes in neurodegenerative diseases.
• Accurate reconstruction of intra-voxel crossing fibers is essential for understanding brain architecture.

Purpose of the Study:

• To propose a novel penalized multi-scale adaptive regression model (PMARM) framework for inferring the orientation distribution function (ODF).
• To enhance spatial and adaptive estimation of water diffusion ODFs in regions with complex fiber configurations.

Main Methods:

• Reformulated HARDI reconstruction as a weighted regularized least-squares regression (WRLSR) problem.
• Incorporated similarity and distance weights for spatial smoothness while preserving discontinuities.
• Utilized L1 penalty for sparse ODF solutions and a scaled L1 weighted estimator for bias correction.
• Integrated multiscale adaptive regression, propagation-separation method, and Lasso for adaptive ODF estimation.

Main Results:

• The PMARM framework demonstrated reduced angle detection errors in fiber crossing areas.
• Achieved more accurate HARDI reconstruction compared to standard voxel-wise methods.
• Successfully inferred ODFs in regions with complex fiber configurations.

Conclusions:

• PMARM offers a robust approach for accurate ODF estimation in HARDI.
• The method enhances the analysis of brain connectivity patterns and their alterations in disease states.
• PMARM provides a valuable tool for advancing neuroimaging research and clinical applications.