Neonatal seizure detection and localization using time-frequency analysis of multichannel EEG

Contrarily to adults and older children, the clinical signs of seizure in newborns are either subtle or occult. For this reason, the electroencephalogram (EEG) has been the most dependable tool used for detecting seizures in newborns. Given nonstationary and multicomponent EEG signals, time-frequency (TF) based methods were found to be very suitable for the analysis of such signals. The TF domain techniques are utilized to extract TF signatures that are characteristic of EEG seizures. In this paper, multichannel EEG signals are processed using a TF matched filter to detect and to geometrically localize neonatal EEG seizures. The threshold used to distinguish between seizure and nonseizure is data-dependent and is set using the EEG background. Multichannel geometrical correlation, based on a concept of incidence matrix, was utilized to further enhance the performance of the detector.