Overlapping brain Community detection using Bayesian tensor decomposition

Biomarkers Derived from Alterations in Overlapping Community Structure of Resting-state Brain Functional Networks for Detecting Alzheimer's Disease

Recent studies show that overlapping community structure is an important feature of the brain functional network. However, alterations in such overlapping community structure in Alzheimer's disease (AD) patients have not been examined yet. In this study, we investigate the overlapping community structure in AD by using resting-state functional magnetic resonance imaging (rs-fMRI) data. The collective sparse symmetric non-negative matrix factorization (cssNMF) is adopted to detect the overlapping community structure. Experimental results on 28 AD patients and 32 normal controls (NCs) from the ADNI2 dataset show that the two groups have remarkable differences in terms of the optimal number of communities, the hierarchy of communities detected at different scales, network functional segregation, and nodal functional diversity. In particular, the frontal-parietal and basal ganglia networks exhibit significant differences between the two groups. A machine learning framework proposed in this paper for AD detection achieved an accuracy of 76.7% when using the detected community strengths of the frontal-parietal and basal ganglia networks only as input features. These findings provide novel insights into the understanding of pathological changes in the brain functional network organization of AD and show the potential of the community structure-related features for AD detection.

Nodal Memberships to Communities of Functional Brain Networks Reveal Functional Flexibility and Individualized Connectome

Human brain network is organized as interconnected communities for supporting cognition and behavior. Despite studies on the nonoverlapping communities of brain network, overlapping community structure and its relationship to brain function remain largely unknown. With this consideration, we employed the Bayesian nonnegative matrix factorization to decompose the functional brain networks constructed from resting-state fMRI data into overlapping communities with interdigitated mapping to functional subnetworks. By examining the heterogeneous nodal membership to communities, we classified nodes into three classes: Most nodes in somatomotor and limbic subnetworks were affiliated with one dominant community and classified as unimodule nodes; most nodes in attention and frontoparietal subnetworks were affiliated with more than two communities and classified as multimodule nodes; and the remaining nodes affiliated with two communities were classified as bimodule nodes. This three-class paradigm was highly reproducible across sessions and subjects. Furthermore, the more likely a node was classified as multimodule node, the more flexible it will be engaged in multiple tasks. Finally, the FC feature vector associated with multimodule nodes could serve as connectome "fingerprinting" to gain high subject discriminability. Together, our findings offer new insights on the flexible spatial overlapping communities that related to task-based functional flexibility and individual connectome "fingerprinting."