Network Dependency Index Stratified Subnetwork Analysis of Functional Connectomes: An application to autism

TL;DR

This study extends the network dependency index (NDI) framework to functional connectomes, demonstrating its stability and utility in identifying group differences in autism spectrum disorder, thereby aiding understanding of brain connectivity alterations.

Contribution

The paper introduces the application of NDI-based subnetwork analysis to functional MRI data and validates its stability and effectiveness in distinguishing ASD-related connectivity differences.

Findings

NDI is feasible on resting-state functional MRI data.

NDI-based subnetworks are stable across different groups.

Significant differences in NDI tiers were found between groups.

Abstract

Autism spectrum disorder (ASD) is a neurodevelopmental condition impacting high-level cognitive processing and social behavior. Recognizing the distributed nature of brain function, neuroscientists are exploiting the connectome to aid with the characterization of this complex disease. The human connectome has demonstrated the brain to be a highly organized system with a centralized core vital for effective function. As such, many have used this topological principle to not only assess core regions, but have stratified the remaining graph into subnetworks depending on their relation to the core. Subnetworks are then utilized to further understand the supporting role of more peripheral nodes with respects to the overall function in the network. A recently proposed framework for subnetwork definition is based on the network dependency index (NDI), a measure of a node's importance based on its contribution to overall efficiency in the network, and the derived subnetworks, or Tiers, have been shown to be largely stable across ages in structural networks. Here, we extend the NDI framework to test its efficacy against a number experimental conditions. We first not only demonstrated NDI's feasibility on resting-state functional MRI data, but also its stability irrespective of the group connectome on which NDI was determined for various edge thresholds. Secondly, by comparing network theory measures of transitivity and efficiency, significant group differences were identified in NDI Tiers of greatest importance. This demonstrates the efficacy of utilizing NDI stratified subnetworks, which can help to improve our understanding of diseases and how they affect overall brain connectivity.