EEG-Based Minimum Spanning Tree Analysis Reveals Network Disruptions in Alzheimer's Disease Spectrum: An Observational Study

Xing Ye¹Yixin Yan²Yingying Wang³Jingping Shi^1*

• ¹Department of Neurology, Nanjing Brain Hospital Affiliated to Nanjing Medical University, Nanjing, China
• ²Department of Neurology, Shanghai Public Health Clinical Center, Fudan University, Shanghai, China
• ³Department of Neurology, Huai'an 82 Hospital, Huai‘an, China

Introduction: Alzheimer's disease (AD) is characterized by disrupted brain connectivity, but the network changes across disease stages remain poorly understood. This observational cross-sectional study investigated alterations in functional brain networks across the AD continuum using minimum spanning tree (MST) analysis of resting-state EEG (rsEEG) data. Methods: We analyzed rsEEG data from 65 participants (30 healthy controls, 14 mild cognitive impairment due to AD [MCI-AD], 21 AD). Phase Lag Index (PLI)-based connectivity and MST metrics (such as diameter, eccentricity, and maximum degree) were computed across five frequency bands. Group differences were assessed using Kruskal-Wallis tests, and correlations with cognitive measures, disease severity, and cerebrospinal fluid (CSF) biomarkers were examined. Results: Significant alterations in rsEEG network topology were observed across HC, MCI-AD, and AD groups. AD patients showed increased theta band connectivity (higher mean PLI, diameter, and eccentricity) and decreased beta band connectivity (lower mean PLI and eccentricity) compared to HC. MCI-AD group exhibited higher delta band maximum degree and altered beta band network organization compared to HC and AD. These network changes correlated with cognitive performance and disease severity. Beta band mean PLI and theta band eccentricity effectively discriminated between AD/MCI-AD and HC. Significant correlations were also found between specific MST metrics and CSF biomarkers (t-Tau, p-Tau, Aβ1-42). Conclusion: AD progression is characterized by frequency-specific alterations in brain network topology, particularly in theta and beta bands, detectable through rsEEG-based MST analysis. These findings suggest EEG-derived network measures may serve as potential biomarkers for early AD diagnosis and monitoring disease progression.