Recurrence based deterministic trends in EEG records of epilepsy patients

Nonlinear EEG analysis has recently received increasing attention in an attempt to detect and predict seizures. This paper explores the deterministic trends in the epilepsypsilas EEG records by using quantitative recurrence methods. Two quantitative indexes, the recurrence rate (RR) and deterministic trends (DET) are introduced to investigate the hidden features in brain electrophysiological records. The analysis on both intracranial and scalp EEG recordings shows that the recurrence rate and the deterministic trends index tend to keep a low value 5-10 minutes before seizure onset. This preliminary study from 2 epilepsy patients suggests that RR and DET maybe useful in identifying a pre-ictal state. Future research will investigate these methods in a larger cohort of patients.