Within-Individual Precision Mapping of Brain Networks Exclusively Using Task Data

Abstract

Precision mapping of brain networks within individuals has become a widely used tool that prevailingly relies on functional connectivity analysis of resting-state data. Here we explored whether networks could be precisely estimated solely using data acquired during active task paradigms. The straightforward strategy involved extracting residualized data after application of a task-based general linear model (GLM) and then applying standard functional connectivity analysis. Functional correlation matrices estimated from task data were highly similar to those derived from traditional resting-state fixation data. The largest factor affecting similarity between correlation matrices was the amount of data. Networks estimated within-individual from task data displayed strong spatial overlap with those estimated from resting-state fixation data and predicted the same triple functional dissociation in independent data. The implications of these findings are that (1) existing task data can be reanalyzed to estimate within-individual network organization, (2) restingstate fixation and task data can be pooled to increase statistical power, and (3) future studies can exclusively acquire task data to both estimate networks and extract task responses. Most broadly, the present results suggest that there is an underlying, stable network architecture that is idiosyncratic to the individual and persists across task states.