Fig. 4: Stages of building frame network.

For each subject, the connectivity matrix was converted into a column vector, and the connectivity strength of each edge was sequenced to get the ranking value. The group average ranking of each edge was then calculated. The coefficient of variation of each edge was obtained (standard deviation divided by the mean). The edges with the lowest coefficients of variation (the top 1%, 41 edges) were selected to construct the frame network of the group.