Brain Transfer: Spectral Analysis of Cortical Surfaces and Functional Maps

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Anatomy

Background:

• Accurate alignment of cortical data is crucial for population-based fMRI studies.
• Current methods for cortical surface inflation, matching, and functional data alignment are often separate, computationally intensive, and rely on potentially variable anatomical features.

Purpose of the Study:

• To introduce BrainTransfer, a unified spectral framework to efficiently process and align cortical data.
• To overcome limitations of existing methods by integrating smoothing, matching, and functional map transfer.
• To enhance the accuracy and speed of group-level fMRI analyses.

Main Methods:

• Developed a spectral framework (BrainTransfer) unifying cortical smoothing, point matching with confidence regions, and functional map transfer.
• Optimized a spectral transformation matrix combining point correspondence and eigenbasis change.
• Introduced focused harmonics to localize spectral decomposition of functional data.

Main Results:

• BrainTransfer performs unified cortical processing, including smoothing, matching, and functional map transfer, within minutes.
• Achieved increased matching accuracy on retinotopy compared to conventional methods.
• Demonstrated benefits through a variability study on shape and functional data, highlighting improved statistical analysis on surfaces.

Conclusions:

• BrainTransfer offers a computationally efficient and accurate spectral solution for aligning cortical data across individuals.
• The framework enables reliable transfer of surface functions and localized confidence assessments.
• Provides a novel approach for direct statistical analysis on cortical surfaces in neuroimaging research.