Prispevek časovne informacije k izračunu možganske funkcijske konektivnosti pri oceni njene povezave z medosebnimi razlikami v vedenju

Functional connectivity (FC) refers to the statistical associations of time series between signals from different brain regions. We investigated the role of temporal information in FC computation for understanding brain-behavior associations and improving the generalizability of such associations. Two aspects of temporal information in FC computation were considered: sensitivity to the temporal order and variability of FC estimates over time. We examined both aspects in two studies. The first study compared FC methods that differ in their sensitivity to the temporal order (static vs. dynamic) and the number of regions considered when estimating a single edge (bivariate vs. multivariate). Similarities between FC matrices, node centrality measures, and patterns of brain-behavior associations were assessed. Although FC estimates did not differ as a function of sensitivity to temporal order, we observed differences between the multivariate and bivariate FC methods. We conclude that dynamic FC estimates represent information that is largely similar to that of static FC. In the second study, we examined the between-subject and within-subject variability of FC estimates. We hypothesized that brain features with higher between-subject variability and lower within-subject variability would be more informative about brain-behavior associations than other features. We showed that FC edges varying across subjects were primarily located in associative networks, whereas FC edges varying across time were located in unimodal networks. Resting-state FC was predominantly influenced by stable individual factors, whereas time-scale variation was smaller and more subject-specific. We used canonical correlation analysis to examine brain-behavior associations. Although models based on edges with low within-subject and high between-subject variability performed similarly to models based on all edges, edges with higher between-subject variability were generally more informative than edges with lower between-subject variability. Our research highlights the importance of carefully considering multiple factors throughout the analysis pipeline when investigating brain-behavior associations.