Coactivated Clique Based Multisource Overlapping Brain Subnetwork Extraction

TL;DR

This paper introduces a novel method for extracting overlapping brain subnetworks by identifying co-activated cliques across multiple data sources, improving reproducibility and neuroscientific relevance over existing techniques.

Contribution

It proposes a multisource co-activated clique approach for overlapping subnetwork extraction, incorporating task information and automatic clique size detection.

Findings

Enhanced reproducibility of subnetwork extraction.

Better alignment with known brain hubs.

Improved identification of overlapping regions.

Abstract

Subnetwork extraction using community detection methods is commonly used to study the brain's modular structure. Recent studies indicated that certain brain regions are known to interact with multiple subnetworks. However, most existing methods are mainly for non-overlapping subnetwork extraction. In this paper, we present an approach for overlapping brain subnetwork extraction using cliques, which we defined as co-activated node groups performing multiple tasks. We proposed a multisource subnetwork extraction approach based on the co-activated clique, which (1) uses task co-activation and task connectivity strength information for clique identification, (2) automatically detects cliques of different sizes having more neuroscientific justifications, and (3) shares the subnetwork membership, derived from a fusion of rest and task data, among the nodes within a clique for overlapping subnetwork extraction. On real data, compared to the commonly used overlapping community detection techniques, we showed that our approach improved subnetwork extraction in terms of group-level and subject-wise reproducibility. We also showed that our multisource approach identified subnetwork overlaps within brain regions that matched well with hubs defined using functional and anatomical information, which enables us to study the interactions between the subnetworks and how hubs play their role in information flow across different subnetworks. We further demonstrated that the assignments of interacting/individual nodes using our approach correspond with the posterior probability derived independently from our multimodal random walker based approach.