Joint modelling of diffusion MRI and microscopy.

Area of Science:

• Neuroimaging
• Biophysics
• Histology

Background:

• Diffusion MRI (dMRI) offers insights into tissue microstructure but relies on indirect signals.
• Microscopy is typically used separately to validate dMRI models, treating it as a gold standard.
• Current dMRI analysis often treats modalities independently, limiting comprehensive understanding.

Purpose of the Study:

• To develop and validate a joint dMRI-histology model for simultaneous analysis of multimodal data.
• To leverage complementary information from dMRI and microscopy for a more integrated understanding of tissue microstructure.
• To address limitations in current dMRI models, such as the degeneracy between fiber dispersion and radial diffusion.

Main Methods:

• A generative joint dMRI-histology model was developed for simultaneous data analysis.
• The framework was applied to spherical-deconvolution analysis to refine microstructure modeling.
• The model investigated the dependency of the fiber response function on local anatomy.

Main Results:

• The joint model successfully integrated dMRI and microscopy data from the same tissue samples.
• The study demonstrated that the fiber response function is anatomically dependent, challenging 'brain-wide' assumptions.
• Results indicate potential overestimation of dispersion and underestimation of fiber populations in current dMRI models.

Conclusions:

• Simultaneous analysis of dMRI and microscopy data within a joint model offers significant advantages over separate analyses.
• The proposed framework enhances the accuracy of diffusion MRI microstructure modeling, particularly in resolving fiber orientation and dispersion.
• This approach provides a more refined understanding of white matter complexity and variability across different brain regions.