Figure 6

Flowchart of community detection algorithm.

The procedures are demonstrated using eight subjects. For each AC obtained in gRAICAR, its similarity matrix (A) is represented using an undirected weighted graph (B), where the red nodes stand for the eight subjects and the distance between nodes indicates the similarities of the spatial maps from different subjects (shorter distance means higher similarity). The graph is then converted into a binary graph (C), where every two nodes are either connected or disconnected, by thresholding the edges with a probability threshold, p, derived from gRAICAR permutation tests. Within this binary graph, an algorithm is performed to search for maximal-cliques. In a maximal-clique, every pair of nodes is connected, indicating a dense cluster formed by the element nodes. In (D), three colors are used to mark three different maximal-cliques. The maximal-cliques are then merged if they share k nodes. A subject community is defined by the merged maximal-cliques (E).