Effects of Tasks on Functional Brain Connectivity Derived from Inter-Individual Correlations: Insights from Regional Homogeneity of Functional MRI Data

Abstract

Research on brain functional connectivity often relies on intra-individual moment-to-moment correlations of functional activity, typically using functional MRI (fMRI). Inter-individual correlations are also employed on data from fMRI and positron emission tomography (PET). Many studies have not specified tasks during scanning, keeping participants in an implicit "resting" condition. This lack of task specificity raises questions about how different tasks impact inter-individual correlation estimates. In our analysis of fMRI data from 100 unrelated participants, scanned during seven tasks and in a resting state, we calculated Regional Homogeneity (ReHo) for each task as a regional measure of brain functions. We found that changes in ReHo due to tasks were relatively small compared with its variations across brain regions. Cross-region variations of ReHo were highly correlated among tasks. Similarly, whole-brain inter-individual correlation patterns were remarkably consistent across the tasks, showing correlations greater than 0.78. Changes in inter-individual correlations between tasks were primarily driven by connectivity in the visual, somatomotor, default mode network, and the interactions between them. This subtle yet statistically significant differences in functional connectivity may be linked to specific brain regions associated with the studied tasks. Future studies should consider task design when exploring inter-individual connectivity in specific brain systems. Impact StatementInter-individual correlation is increasingly used to estimate brain connectivity, complementing intra-individual correlations in fMRI, particularly for measures like cerebral blood flow obtained via fMRI and PET. However, how task performance affects inter-individual correlations is largely unknown. This study used regional homogeneity as a summary measure of brain functions from task-based fMRI data across eight tasks. The inter-individual correlations were highly similar across tasks, indicating the underlying brain network structure can be inferred under various conditions. Subtle but statistically significant differences in connectivity estimates suggest the functional significance of this approach.