Classification of wakefulness and anesthetic sedation using combination feature of EEG and ECG

Abstract

There have been lots of trials to classify a depth of anesthesia using diverse physiological indices. In this study, we classified wakefulness and propofol-induced sedation using combined electroencephalography (EEG) and electrocardiography (ECG) features for better classification performance. We extract each spectral band of EEG and very low frequency (VLF) of heart rate variability using spectrogram and low-pass filter, respectively. We used combined feature of EEG spectral bands and VLF and shrinkage-regularized linear discriminant analysis as a classifier. Our results show that combination of EEG spectral power and VLF can improve the classification performance between wakefulness and sedation from 95.1±5.3% to 96.4±4.2%.