Sleep spindles and spike–wave discharges in EEG: Their generic features, similarities and distinctions disclosed with Fourier transform and continuous wavelet analysis

Abstract

Epileptic activity in the form of spike–wave discharges (SWD) appears in the electroencephalogram (EEG) during absence seizures. A relationship between SWD and normal sleep spindles is often assumed. This study compares time–frequency parameters of SWD and sleep spindles as recorded in the EEG in the WAG/Rij rat model of absence epilepsy. Fast Fourier transformation and continuous wavelet transformation were used for EEG analysis. Wavelet analysis was performed in non-segmented full-length EEG. A specific wavelet-based algorithm was developed for the automatic identification of sleep spindles and SWD. None of standard wavelet templates provided precise identification of all sleep spindles and SWD in the EEG and different wavelet templates were imperative in order to accomplish this task. SWD were identified with high probability using standard Morlet wavelet, but sleep spindles were identified using two types of customized adoptive ‘spindle wavelets’. It was found that (1) almost 100% of SWD (but only 50–60% of spindles) were identified using the Morlet-based wavelet transform. (2) 82–91% of sleep spindles were selected using adoptive ‘spindle wavelet 1’ (template's peak frequency ∼12.2 Hz), the remaining sleep spindles with ‘spindle wavelet 2’ (peak frequency ∼20–25 Hz). (3) Sleep spindles and SWD were detected by the elevation of wavelet energy in different frequencies: SWD, in 30–50 Hz band, sleep spindles, in 7–14 Hz. It is concluded that the EEG patterns of sleep spindles and SWD belong to different families of phasic EEG events with different time frequency characteristics.