Personalized preictal EEG pattern characterization: do timing and localization matter?

Objectives: Better understanding of ictogenesis may allow clinical interventions and potentially reduce the impact of epilepsy on patients' quality of life. This study aims to characterize the EEG changes during the preictal period. Methods: This work retrospectively analyzed long-term scalp EEG recordings collected at two neurology centers to characterize preictal activity (start point and duration) for each seizure using EEG features. A channel selection algorithm was implemented and localized preictal activity. Results: Out of 19 patients, 17 (89.5%) had a distinct preictal pattern, starting 83 ± 60 min before seizure onset and lasting 56 ± 47 min. Spectral Entropy and Hjorth mobility were consistently two out of the three features best distinguished preictal from interictal activity. The third distinguishing feature was either theta power, delta power, beta power, or gamma power. Preictal activity before two seizures in the same patient shared common electrodes and features but differed in duration and timing. Conclusion: Preictal activity, defined as prolonged intervals of uncommon EEG activity, varies in time, localization and signal patterns between individuals and varies in timing and duration between seizures of the same individual.