A2053
Title: Overview of methods for characterizing brain functional connectivity
Authors: Hernando Ombao - KAUST (Saudi Arabia) [presenting]
Abstract: Modeling dependence between nodes in a brain network is central to understanding underlying neural mechanisms. A broad range of statistical methods for characterizing dependence in a brain network is presented. An overview of some of the most prominent frequency domain measures, such as coherence, partial coherence, and dual-frequency coherence is provided. This is followed by recent developments. First, wavelet partial canonical coherence (WPCanCoh) derived under multivariate locally stationary wavelet processes is discussed. Second, current work on information-theoretic measures of connectivity including the spectral transfer entropy (STE) and the non-linear vector coherence (NVC), which quantify the magnitude and direction of information flow from a certain frequency-band oscillation of a channel to an oscillation of another channel, is briefly described. In the final part, the use of deep learning techniques to estimate high-dimensional nonparametric models that explain how activity in one channel can be explained by potential non-linear past activity from other channels in the network is explored. These methods are illustrated through numerical experiments and provide interesting and novel findings on the analysis of EEG recordings from different cognitive tasks.
