Using high angular resolution diffusion imaging data to discriminate cortical regions.

Area of Science:

• Neuroscience
• Neuroimaging
• Biomedical Engineering

Background:

• Brodmann's map, a century-old standard for cortical localization, requires updating with modern, high-resolution, and individualized methods.
• Existing methods lack the ability to capture individual variability in cortical structure.
• There is a need for non-invasive techniques to map grey matter microstructure.

Purpose of the Study:

• To demonstrate that standard High-angular resolution diffusion imaging (HARDI) data can be used to update cortical parcellation.
• To establish HARDI's potential for creating individualized, cortex-wide maps of grey matter microstructure.
• To assess the reproducibility of HARDI-based parcellation across scans.

Main Methods:

• Identified gray/white matter and pial boundaries using high-resolution structural MRI.
• Collected two HARDI datasets per individual, aligned with structural images.
• Extracted diffusion-weighted signal at each vertex, deriving orientationally invariant features as a tissue fingerprint.
• Utilized a support-vector machine classifier to distinguish cortical areas based on HARDI features.

Main Results:

• HARDI data exhibits sufficient directional variation in grey matter for functional region parcellation.
• This variation is reproducible across multiple scans.
• A support-vector machine classifier achieved 80-82% accuracy in classifying distinct cortical areas using HARDI features.
• Demonstrated the potential for creating an individual's cortex-wide map of grey matter microstructure.

Conclusions:

• HARDI-based signal variation is non-random and reproducible, crucial for successful cortical parcellation.
• This approach offers a promising foundation for a new, non-invasive cortical parcellation method in living humans.
• Highlights the potential of grey matter anisotropy, previously overlooked, for neuroimaging applications.