Fig. 1
Detecting multiple functional connectivity states using a sliding-window approach. a The functional network organization of the human cerebral cortex is revealed through intrinsic functional connectivity. Colors reflect regions estimated to be within the same network determined based on the 17-network solution from Yeo et al.39. The map is displayed for multiple views of the left hemisphere in Caret PALS space79. b Correlation matrices are computed across regions from windowed portions (width = 33 s) of each participant’s component time series (n = 1919), and aggregated across the full sample. c K-means clustering was applied to identify repeated patterns of connectivity (brain states). d Instability of the clustering algorithm is plotted as a function of the number of estimated states (2–20). As expected based on increasing solution space (complexity), instability was greater with increasing the number of estimated states per solution. The local minima of the graphs observable at the 2-, 4-, 5-, and 8-state solutions indicate the number of states that can be stably estimated by the selected clustering algorithm with the present data. Resampling over time (sliding windows) and across space (regions of interest) yields comparable results (see Supplementary Figure 1). In this article, we focus on state solutions 2 through 8 to provide a broad survey of the solution space
