Noise-assisted intrinsic mode function coherence in seizure anticipation

Epilepsy is a common neurological disorder characterized by recurrent electrophysiological activities, known as seizures. We explore the applicability of noise-assisted Ensemble Empirical Mode Decomposition (EEMD) for patient-specific seizure anticipation. Intracranial EEG data were obtained from invasive pre-surgical epilepsy monitoring at the Epilepsy Center of the University Hospital of Freiburg. Data from six patients (19 seizure recordings) with hippocampal foci were analyzed. For each recorded channel, twelve levels of intrinsic mode functions (IMFs) were produced. The coherence between the IMFs (denoted as IMF-Coh) between different channel pairs was computed. Statistical distributions of IMF coherence were determined from three hours of interictal data. Patient-, IMF level-, and channel pair-specific IMF-Coh were used to determine the earliest anticipation times for detected ictal events. Our study shows that while not all channel pairs are able to detect every ictal event, in general, low IMFs (containing frequency components greater than 1 Hz) can discriminate between interictal and periictal activities. Our results suggest patient-specific increases in coherence for one or more IMF levels during seizure progression. The anticipation window ranges from 30 to 53 minutes prior to clinical manifestation. We propose an anticipation optimality index as a joint indicator of sensitivity and earliest anticipation times to help select relevant channel pairs and IMF levels. In future work, we will incorporate cross-validation techniques with more interictal data as well as investigate patient-specific, automated selection of high-sensitivity channel pairs.