Functional parcellation of memory related brain networks by spectral clustering of EEG data

In this study, we investigate the clustering information of alpha band brain networks during memory load task. For this purpose, short time memory task which includes memory load varieties is implemented to the subjects. To calculate mutual information, time and frequency information is both taken into consideration due to Cohen class time-frequency distribution (TFD) formulation. Cohen class based mutual information helps us to integrate adjacency matrices based on the similarity information of individual electrode pairs. In addition, essential frequency bins are selected from the TFD with respect to the default alpha frequency (8 - 12Hz) intervals. Moreover, graph based spectral clustering algorithm is used to parcellate memory related circuits on the brain. From the calculated adjacency matrices, the N-cut algorithm is used for node wise clustering between nodes. After node wise clustering information, subject wise clustering is applied with respect to the similarities of node information over all subjects.