Graph-Theoretic Functional Connectivity Analysis of EEG for Discrimination of Interictal, Preictal, and letal States in Epileptic Seizure Dynamics

Abstract

Epileptic seizures are associated with complex and dynamic alterations in functional brain connectivity. In this study, a graph-theoretic framework is proposed to analyze Scalp EEG signals for distinguishing interictal, preictal, and ictal states. Functional connectivity is computed using Phase Locking Value (PLV), Phase Lag Index (PLI), and Pearson correlation, capturing both phase- and amplitude-based interactions between EEG channels. These connectivity matrices are transformed into graph representations, from which network-level features such as node degree, clustering coefficient, betweenness centrality, and global efficiency are extracted. Statistical evaluation using one-way ANOV A and Kruskal-Wallis tests reveals highly significant differences across all three seizure states for all connectivity-derived features, with p-values consistently below 0.001. Tue results demonstrate that epileptic seizures are associated with pronounced reorganization ofbrain functional networks, where preictal states exhibit intermediate connectivity characteristics between interictal and ictal conditions. Tue proposed framework provides an interpretable and physiologically meaningful approach for seizure state characterization using EEG-based network analysis.