Time-frequency analysis of high-frequency activity for seizure detection and tracking in neonate

Time-frequency based methods have been shown to outper-form other methods in dealing with newborn EEG. This is due to the fact that newborn EEG is nonstationary and multicomponent. This paper presents a new time-frequency based EEG seizure detection method. It uses the distribution of the interspike intervals of a high frequency slice of the time-frequency representation of an EEG epoch to discriminate between seizure and non-seizure activities. The seizure detected through this method is then tracked throughout all the available EEG channels by cross-correlating the binary encoded signals of both the detected seizure and the subsequent EEG epochs in all channels. This approach allows the study of the migrating behavior of seizure using EEG signals.

2. Introduction

Neonatal seizure is a common occurrence in newborn. Approximately one in every 200 newborn babies experiences some forms of seizure, indicating cerebral abnormalities or damage to the brain.