Detection of epileptic seizures in EEG signals during long-term monitoring of patients after traumatic brain injury

Abstract

Long-term (several days) monitoring of epileptiform activity in scalp EEG of posttraumatic brain injury patients is an important task. EEG signals contain epileptiform seizures and similar signals of myographic activity associated with chewing. Both epileptiform activity and chewing artifacts appear in the same frequency range, which complicates their differentiation. To distinguish epileptiform activity from chewing artifacts, a method based on the wavelet spectrogram analysis of EEG is proposed. EEG wavelet spectrogram contains broadband peaks at times corresponding to peak-wave epileptiform activity on the one hand, and peaks of myographic activity at chewing on the other hand. The periodicity of these peaks is investigated. The difference in the period dispersion of epileptiform peaks and chewing peaks are found.