Hypergraph based Subnetwork Extraction using Fusion of Task and Rest Functional Connectivity

TL;DR

This paper introduces a hypergraph-based method that combines multi-task and resting state fMRI data to improve brain subnetwork extraction, addressing noise issues and capturing higher-order relations.

Contribution

It presents a novel high order relation informed hypergraph approach that integrates multiple data sources for more accurate and biologically meaningful subnetwork identification.

Findings

Improved subnetwork modularity compared to single-source methods.

Higher inter-subject reproducibility of extracted subnetworks.

More biologically meaningful network assignments.

Abstract

Functional subnetwork extraction is commonly used to explore the brain's modular structure. However, reliable subnetwork extraction from functional magnetic resonance imaging (fMRI) data remains challenging due to the pronounced noise in neuroimaging data. In this paper, we proposed a high order relation informed approach based on hypergraph to combine the information from multi-task data and resting state data to improve subnetwork extraction. Our assumption is that task data can be beneficial for the subnetwork extraction process, since the repeatedly activated nodes involved in diverse tasks might be the canonical network components which comprise pre-existing repertoires of resting state subnetworks. Our proposed high order relation informed subnetwork extraction based on a strength information embedded hypergraph, (1) facilitates the multisource integration for subnetwork extraction, (2) utilizes information on relationships and changes between the nodes across different tasks, and (3) enables the study on higher order relations among brain network nodes. On real data, we demonstrated that fusing task activation, task-induced connectivity and resting state functional connectivity based on hypergraphs improves subnetwork extraction compared to employing a single source from either rest or task data in terms of subnetwork modularity measure, inter-subject reproducibility, along with more biologically meaningful subnetwork assignments.