A Tucker Decomposition Based Approach For Topographic Functional Connectivity State Summarization

The brain reconfigures itself continuously in response to different external stimuli. Advances in noninvasive brain activity recording has made it possible to gain insight into the functional brain activity over time. The functional connectivity has been mostly characterized as a static network through linear and nonlinear measures of statistical dependency. However, recent work indicates that functional connectivity is dynamic and this dynamic reconfiguration of connections accounts for various cognitive functions. The goal of this study is to provide a concise summarization of the quasi-stationary functional connectivity network state within kl time interval across subjects. We propose to consider the functional connectivity networks constructed by bivariate phase synchrony measure as tensors and use Tucker decomposition to obtain a low rank approximation to summarize the network. The significant connections within ki given network state are obtained through significance testing. Finally, the proposed framework is applied to multichannel electroencephalogram (EEG) data front a study of error processing in the brain to investigate the connectivity patterns during error and correct responses.