NEUROPHYSIOLOGY AND CONNECTOMICS: TOWARD A BRAIN FUNCTION ATLAS

Abstract

To portray brain organisation at multiple levels, the complete atlas of brain functions need to combine neurophysiological, neuroimaging, as well as computational studies.  The study utilized high-density electrophysiological recordings, local field potential (LFP) spectral analysis, functional MRI (fMRI) activation maps and diffusion MRI connectomics in a mixed-method study.  The high-degree hubs of a modular small-world architecture were identified in associative cortices graph-theoretically.  Whereas in task-based analyses there was a significant rewiring of the connections, the resting-state networks indicated many connections within connection pattern.  Cross-frequency linkages within particular regions have been identified and related to number cross-frequency links was indicated in a study of phase-amplitude coupled (PAC) in which oscillatory coordination was associated with network-integration.  The performance of the models formed based on structural and functional parameters was verified when machine learning models such as XGBoost and neural networks had classification accuracies greater than 95% in functional network identification.  Integration of these data resulted in a multi-scale, multi-modal description of brain activity bringing the field closer to its goal of an comprehensive atlas of brain activity.  The framework can form the foundation of therapeutic applications, including personalised mapping in neurological and psychiatric illnesses and the basic neuroscience.