Figure 1.

Illustration of the probabilistic model. A, Dynamic functional connectivity matrices at different sliding windows for one participant (here using the AAL-90 atlas). The axes denote ROIs, which are listed in Table 1. The color bar denotes the Pearson correlation coefficients. B, The functional connectivity patterns for a given ROI i at different sliding windows. We used a local thresholding method that reserves its k (k = 5 shown here) strongest functional connections (red lines) at each time window. C, The probability distribution P_i(j … n) reflects the frequency of each connection with i emerged across the temporal windows. A complexity measure (i.e., entropy H_i) was then applied to this probability distribution to quantify functional flexibility of region i.