Automatic classification of sleep staging by electroencephalogram (EEG) is one of the problems that has remained in the field of biomedical engineering. The medical doctor or the qualified electroencephalographer takes much time and work to execute the sleep staging by the recorded papers of the sleeping EEG.
In this study, a method for automatic classification of the sleep staging based on the power spectrum features of the EEG, and on the EOG and the EMG has been developed. First, the features of the EEG on the frequency domain are extracted by applying the autoregressive model to the sleeping EEG. The features are characterized in terms of the coefficients of the autoregressive model. Then distances of a maximum likelihood statistic are computed by using the coefficients.Thus the characteristic waves are determined on the sleeping EEG in due order. Next, from the characteristic waves and information on the eye movement as well as on the activity of the EMG, the sleep staging is carried out. This staging is based on the sleep stage manual by Rechtschaffen and Kales. Finally, these results calculated by the computer are compared with those of a medical doctor. These techniques developed here will be applicable to the anesthesia level EEG classification.