Model-identified excitation/inhibition changes differentiate seizure onset patterns in intracranial recordings of epilepsy patients

Abstrakt

PURPOSE: Current research in epilepsy focuses on unraveling the heterogeneity among seizure phenotypes to enhance our understanding of epilepsy, increase diagnostic accuracy, and improve treatment outcomes. Computational models offer mechanistic insight into different types of seizure transitions and, in the future, could be used as diagnostic tools in the context of individual treatment planning, for example when applied to the intracranial EEG data recorded during presurgical evaluation (Dallmer-Zerbe et al, Epilepsia 2023; 64(9): 2221–2238). QUESTION: In this study, we investigate seizure recordings from 15 patients with drug-resistant epilepsy from the European Epilepsy Database, subdivided into seven clinically marked seizure onset patterns. We ask whether model-identified excitation/inhibition changes, as automatically identified from iEEG data, significantly differ between these different types of seizures. METHOD: Building on the modeling work of Wendling et al. (J Clin Neurophysiol 2005; 22(5): 343–356), we fit simulated and real data segments of five-second length via adjusting the model’s excitation and inhibition parameters and comparing sets of data features from 205 seizures. We investigate excitation/inhibition changes in the time interval from 60 seconds before until 25 seconds after clinically marked seizure onset, averaged across four time windows (interictal, preonset, onset, ictal). RESULT: We find that the identified excitation/inhibition changes could classify the different seizure types significantly above chance (p < .01). Interestingly, the types could be distinguished not only during onset and in clinically marked onset zones, but already in the interictal interval (60–30 seconds before seizure) and in non-onset channels (p < .01). While across types both excitation and inhibition clearly increased during seizure transition, seizure types clearly differed in magnitude and timing of change. CONCLUSION: Different seizure types are linked to distinct onset mechanisms that are pre-determined at least tens of seconds before seizure onset and not (only) localized in the seizure onset zone.