Eeg cortical connectivity analysis of working memory reveals topological reorganization in theta and alpha bands

• Main Authors: Dai, Zhongxiang, de Souza, Joshua, Lim, Julian, Ho, Paul M., Chen, Yu, Li, Junhua, Thakor, Nitish, Bezerianos, Anastasios, Sun, Yu
• Other Authors: School of Computer Science and Engineering
• Format: Article
• Language: English
• Published: 2018
• Subjects: Cortical Functional Connectivity; EEG
• Online Access: https://hdl.handle.net/10356/86208; http://hdl.handle.net/10220/45379
• Institution: Nanyang Technological University

Numerous studies have revealed various working memory (WM)-related brain activities that originate from various cortical regions and oscillate at different frequencies. However, multi-frequency band analysis of the brain network in WM in the cortical space remains largely unexplored. In this study, we employed a graph theoretical framework to characterize the topological properties of the brain functional network in the theta and alpha frequency bands during WM tasks. Twenty-eight subjects performed visual n-back tasks at two difficulty levels, i.e., 0-back (control task) and 2-back (WM task). After preprocessing, Electroencephalogram (EEG) signals were projected into the source space and 80 cortical brain regions were selected for further analysis. Subsequently, the theta- and alpha-band networks were constructed by calculating the Pearson correlation coefficients between the power series (obtained by concatenating the power values of all epochs in each session) of all pairs of brain regions. Graph theoretical approaches were then employed to estimate the topological properties of the brain networks at different WM tasks. We found higher functional integration in the theta band and lower functional segregation in the alpha band in the WM task compared with the control task. Moreover, compared to the 0-back task, altered regional centrality was revealed in the 2-back task in various brain regions that mainly resided in the frontal, temporal and occipital lobes, with distinct presentations in the theta and alpha bands. In addition, significant negative correlations were found between the reaction time with the average path length of the theta-band network and the local clustering of the alpha-band network, which demonstrates the potential for using the brain network metrics as biomarkers for predicting the task performance during WM tasks.