Topological Signatures of Altered Brain Network Centrality in ADHD: A TDA Mapper Study

Ali Nabi Duman

Keywords: Topological Data Analysis, Mapper Algorithm, ADHD, Functional Connectivity, Brain Dynamics, Network Centrality, fMRI

TL;DR: This study uses TDA:MAPPer on fMRI data to demonstrate that ADHD is characterized by more centralized and rigid brain network dynamics, a finding that correlates with symptom severity.

Abstract: Attention-Deficit/Hyperactivity Disorder (ADHD) is a prevalent neurodevelopmental disorder widely hypothesized to stem from alterations in large-scale brain connectivity. However, neuroimaging studies have yielded inconsistent findings, motivating the need for advanced analytical methods capable of capturing the complex, dynamic nature of brain function. In this study, we apply Topological Data Analysis (TDA), specifically the Mapper algorithm, to resting-state functional magnetic resonance imaging (fMRI) data from the multi-site ADHD-200 dataset. We constructed graphical representations of brain state dynamics for participants with ADHD and typically developing controls (TDC) from three independent sites. The topological structure of these graphs was quantified using network centrality measures (betweenness, closeness, and degree). Our results reveal a significant increase in centrality measures in the ADHD group compared to TDC in three cohorts. Furthermore, we observed a weak but significant positive correlation between centrality and symptom severity in one of the cohorts. We conclude that TDA-derived centrality measures can detect alterations in the dynamical organization of brain activity in ADHD, potentially reflecting a less efficient or more rigid network topology.

Submission Number: 54