Time-Frequency Analysis of Human Sleep EEG

We have developed the automatic discrimination system of human sleep EEG (electroencephalogram) stages based on a wave-shape recognition method. These systems can detect discrete stages (stage MT, W, 1, 2, 3, 4 and REM). But, more detailed information extraction is impossible by them. Therefore, we tried to continuous wavelet analysis applied to EEG signals in order to extract more precise information for the stages. And we have proposed a modified wavelet transform method and an extraction method of time series of peak frequency based on time-frequency analysis. It is confirmed that the modified wavelet transform method is effective in some subjects, and it is clarified that feature extraction will be improved by the parameter tuning. In this paper, we analyze the data of six all-night sleep experiments in our laboratory, and tested the effectiveness of the modified wavelet transform method in more detail by tuning the parameters. From these results, we could direct the way of improving the modified wavelet transform method and extracting the feature about sleep stages