Figure 2.

Classification of resting-state fcMRI networks. Classifier output for the sensorimotor network (SMN) and language network (LAN) are shown above in blue (A) and orange (B), respectively. Values were rescaled to a percentile of across all voxels and a threshold at the 80th percentile is illustrated after interpolating to the reconstructed anatomy. Classifier values were sampled to electrodes, which are filled with red to indicate that nearby (up to 5 mm) gray matter voxels were, on average, in the 80th percentile or higher. White electrodes fell below this threshold. A winner-take-all (WTA) summary across all networks (C) was computed and displayed in similar fashion with respective coloring. Each voxel was assigned a color based on the network with the highest class-score. The resulting map was interpolated to the subject’s surface anatomy and to each electrode. The networks illustrated include the default mode (DMN), language (LAN), fronto-parietal control (FPC), visual (VIS), somatomotor network (SMN), ventral attention (VAN), and dorsal attention (DAN). N/A indicates that no network achieved a classifier score above threshold.