A Mass Multivariate Edge-wise Approach for Combining Multiple Connectomes to Improve the Detection of Group Differences.

Functional connectivity derived from functional magnetic resonance imaging data has been extensively used to characterize individual and group differences. While these connectomes have traditionally been constructed using resting-state data, recent work has highlighted the importance of combining multiple task connectomes, particularly for identifying individual differences. Yet, these methods have not yet been extended to investigate differences at the group level. Here, we propose a mass multivariate edge-wise approach to improve the detection of group differences by combining connectomes from multiple sources. For each edge, the magnitude of connection strength from each of multiple connectomes are included in statistical hypothesis testing. We evaluate the proposed approach by estimating sex differences in two large, publicly available datasets: the Human Connectome Project and Philadelphia Neurodevelopmental Cohort. Results indicate the proposed mass multivariate edge-wise analysis offers improved detection of group differences compared to univariate analysis, and support the utility of combining multiple connectomes to improve detection of group differences.