Fig. 1.

Analysis pipeline overview. (a) Pupillometry-based fMRI state stratification for arousal level-dependent connector hub analysis. For each subject, pupillometry data were used to stratify the simultaneously acquired fMRI data into two states (high and low arousal). Specifically, time-points where pupil area was within the top or bottom 20% rank were assigned to a high- (orange) or low-arousal state (blue), respectively. A sparsity-based analysis of reliable k-hubness (SPARK) was used to identify connector hubs from state-stratified fMRI data, by measuring k-hubness for each node at the individual level. (b) k-hubness is defined as the number of overlapping networks in each node. (c) Null data generation by randomizing the assignment of pupillometry to fMRI across the 27 subjects. (d) The distribution of Pearson’s correlation coefficients measured between individual pupillometry time-courses.