Neural field modeling and analysis of large-scale brain dynamics

Abstract

The human brain exhibits complex spatiotemporal dynamics at multiple scales. On length scales of a few tenths of a millimeter to the entire brain, neural field theory (NFT) is a well-established physiologically based model of brain activity that has reproduced many key-features of large-scale brain activity such as those measured with electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). Within this framework, brain activity can be decomposed using modal analysis which uses the natural modes of oscillations, eigenmodes, of neural activity on the cortical surface as the fundamental building blocks for all activity. This thesis investigates the spatiotemporal dynamics of such eigenmodes and how they can be used to predict and analyze brain activity. The spatiotemporal properties of the nine lowest-order eigenmodes are derived and their relationship to cortical geometry is explored. It is shown that eigenmode beating gives rise to complex wave dynamics (including standing, traveling, and rotating waves) which have been observed experimentally for decades. NFT is also used to explain and analyze experimental observations of large-scale brain dynamics from two distinct areas of neuroscience. Firstly, the phenomenon of perceptual echo, whereby random input stimuli at one location are correlated with EEG responses at other locations, is predicted and analysed. Secondly, inverse modeling of EEG data over the sleep-wake cycle is performed on patients with mild cognitive impairment and Alzheimer's disease in order to infer abnormal underlying physiology in these populations. Abnormalities in corticocortical, corticothalamic, and intrathalamic networks are analyzed with reference to known sleep impairments in these clinical populations and hypothesized mechanisms of sleep disruption. The potential of certain model parameters to serve as biomarkers of disease progression is discussed.