Dual tree complex wavelet transform for sleep state identification from single channel electroencephalogram

This work analyzes the suitability of spectral features in the Dual Tree Complex Wavelet Transform (DT-CWT) domain for EEG signal analysis by propounding a DT-CWT based feature extraction scheme. Unlike discrete wavelet transform-DT-CWT ensures limited redundancy and provides approximate shift invariance. To demonstrate the efficacy of DT-CWT for EEG signal analysis, it is applied in conjunction with spectral features to devise a feature extraction scheme for automated sleep staging from single-channel EEG. Our findings suggest that spectral features can distinguish between various sleep stages quite well. The p-values obtained by one-way analysis of variance (AN0VA) and graphical analyses also corroborate with this fact Thus, spectral features in the DT-CWT domain may be used to characterize EEG signal. Furthermore, this work can assist the sleep research community to implement various classification models to put computer-aided sleep scoring into clinical practice.