An fMRI-Based Centrality Analysis of Brain Connectivity in Autism Spectrum Disorder

Recent studies have probed Autism Spectral Disorder's (ASD's) neural substrates, analyzing brain connectivity patterns through techniques like centrality using the resting-state functional Magnetic Resonance imaging (rs-fMRI) data, revealing deficits in social communication, repetitive behaviors, and sensory processing. This research aims to observe alterations in information flow and communication networks by assessing the centrality of key brain regions, distinguishing individuals with ASD from neurotypical controls. Evaluating centrality measures' accuracy using a Graph Attention Network (GAT) model classifier and model interpretability with the attention score visualization unveils significant contributors to the model's performance, providing insights into ASD's neural mechanisms and enhancing our understanding of brain connectivity in this disorder.