To construct a functional brain network, we employed an automated anatomical labeling (AAL) atlas [32] to parcellate the whole brain into 90 regions (45 in each hemisphere, N = 90). The names of the 90 regions and their corresponding abbreviations are listed in Table S1. The mean time series of each region was then acquired by averaging the time series of all voxels within that region. Several sources of spurious variances arising from the estimated head-motion profiles, white matter signals, and whole brain signals were further removed by multiple linear regression analysis [33], [34]. The residual of this regression was then used to substitute for the raw mean time series of the corresponding regions. The functional connectivity between a pair of regions was defined as the Pearson's correlation coefficient in the residual time courses. Thus, a functional connectivity matrix (or correlation matrix) (rij, N×N) can be obtained for each subject. Each functional connectivity matrix can be converted to a binary, undirected network G using a cost threshold (t, 0<t<1), which is equivalent to the ratio between the number of edges
