抄録
A method for automatic detection of vigilance level was developed based on EEG data only, as for pre-processing of automatic EEG interpretation (previous work by the authors). As EEG records during sleep or drowsy condition are quite different from those of wakefulness, drowsy situation of subjects during the course of EEG recording should be strongly avoided. Clinically, however, drowsy EEG data are often obtained despite the recorders' efforts. The detection of drowsiness from EEG data is the most important task for EEGers in interpreting awake EEGs. The task of detecting vigilance level was automatized in this study. The equations for detecting vigilance level were determined so that they would conform to the procedure that a EEGer (one of the authors, H. S.) usually adopts for visual inspection of the actual EEG record, and all threshold values in the equations were determined. The proposed automatic detection method was applied to EEG data of 32 subjects and gave the correct judgement as to the wakefulness in all EEG data. The combined use of the proposed automatic method of the vigilance detection and the automatic EEG interpretation method will be effectively applicable to the clinical use.
