Unveiling Epileptic Networks: Motif Synchronization for Interictal Spike Localization.

Epilepsy affects 50 million people worldwide. Despite advancements in treatment, 30% of patients do not achieve seizure freedom with Anti-epilepsy Drugs, necessitating alternative interventions such as surgical re-section. One of the diagnostic tools for epilepsy is the visual inspection of scalp electroencephalogram (EEG) recordings, a time-consuming process that relies on identifying interictal epileptiform discharges (IEDs). This study proposes a novel IED detection method based on Motif Synchronization (MS), a functional connectivity technique that constructs time-varying graphs (TVGs) from EEG data. MS offers high temporal resolution and reduces spurious connections caused by volume conduction, making it suitable for identifying IEDs and localizing probable epileptogenic foci. EEG data from 10 individuals with mesial temporal lobe epilepsy (taken during EEG-fMRI sessions) were used to detect IEDs and assess the impact of different EEG referencing schemes. TVGs generated by MS showed increased connectivity during IEDs, with median referencing providing the most accurate localization of epileptic regions. The aggregated static networks derived from TVGs revealed highly connected clusters corresponding to IED-generating regions in 9 out of 10 subjects. Additionally, global clustering coefficients were higher during IED segments compared to non-IED segments. This study establishes MS as a computationally efficient and interpretable tool for detecting and localizing IEDs. Future work will expand the analysis to include more patients and explore the integration of MS-based TVGs with graph neural networks for enhanced classification of epileptic activity.