Graph theoretical analysis was used to quantify network properties (Bullmore and Sporns, 2009). The network consisted of nodes (29 electrodes in this study) and edges (SL value between channels), and was characterized by means of network integration, segregation, and nodal importance. Network integration refers to the interactions among specialized brain regions, and represents the ability to combine the information from distributed areas. The path length between brain regions has been used to define the functional integration of the brain (Rubinov and Sporns, 2010). In this study, characteristic path length (L and λ; the mean shortest path length between all nodes) and global/local efficiency (Eg and El) were computed as measures of network integration. Network segregation refers to the existence of specialized functional or anatomical regions within a network. The measures of segregation detect the presence of such regions (i.e., cluster) within a network and the presence of network clusters indicates segregated functional dependencies in the brain. We selected the clustering coefficient (C and γ; the fraction of neighboring nodes being also nodes each other) as a measure of network segregation.
