Modeling the orientation distribution function by mixtures of angular central Gaussian distributions.

Area of Science:

• Medical Imaging
• Neuroscience
• Computational Biology

Background:

• Diffusion weighted imaging (DWI) is crucial for visualizing white matter architecture.
• Current models may have limitations in accurately representing complex fiber orientations within voxels.
• Advanced modeling is needed to improve the resolution and interpretation of DWI data.

Purpose of the Study:

• To develop and validate a novel tensor mixture model for diffusion weighted imaging data.
• To implement an automatic model order selection for determining tensor components.
• To explore the potential of the model for improved imaging of cerebral fiber tracts and identifying clinical markers.

Main Methods:

• Development of a tensor mixture model for DWI data.
• Utilizing an automatic model order selection criterion for voxel-wise tensor components.
• Expansion of the weighted orientation distribution function into angular central Gaussian distributions.
• Validation through extensive simulations and high angular resolution human brain imaging.

Main Results:

• The proposed tensor mixture model effectively images cerebral fiber tracts with improved clarity.
• The model's weighted orientation distribution function can be represented as a mixture of angular central Gaussian distributions.
• Inference on model parameters shows potential for novel clinical markers of altered white matter.

Conclusions:

• The developed tensor mixture model enhances the imaging of white matter in diffusion weighted MRI.
• This approach may lead to new biomarkers for diagnosing and monitoring white matter diseases.
• Accessible R software is provided for computing the tensor mixture model from DWI data.