The Research Of The Complexity Of Sleep EEG And Its Automatic Sleep Stage Scoring

The EEG1generated by the electrical activity of cortical neurons, which contains a lot of information of a human. Among them, the sleep EEG is reflected in the stages of sleep EEG changes. About1/3of lifetime is in sleep stage, so sleep is an important physiological activities. Sleep EEG can not only reflect the quality of sleep, and can reflect the health status of the human body. The study of sleep EEG are of great significance in the theory and practical clinical applications.The EEG is a non-stationary time series, how to analyze the random sequence, and extract the information form EEG is a popular issue. In the past, the study of time series were focused on the issue of prediction, but for these special time series, predict the next value that does not mean much, we are more concerned about the macro feature of the EEG. The complexity measure is to provide a quantitative description of the regularities of changes in the random sequence, and the complexity of time series is one of the more popular indicators in analysis of EEG. In the second chapter, we introduce several methods to measure the complexity of time series, and we simulate and evaluate these methods using the actual EEG data.In fact, Kolmogorov (1965) proposed the concept of relative complexity. From the relative complexity point of view, in the third chapter, we modeled the LZ methods propose a method to measure the relative complexity, and achieved good results using real EEG data test. The relative complexity performance better than complexity when classify two pieces of data.Secondly, in different sleep periods, the waveform characteristics of EEG are quite obvious. M. Hanaoka, etc.(2000) proposed the idea of detecting the characteristic waveform of EEG, but the algorithm is not given. In Chapter IV, we propose a waveform detection method, which is mainly used for the detection of sleep spindles wave and slow wave. Then according to the complexity, quantile, the waveform detection method and the relative complexity, we classify the sleep data. In addition to REM sleep and S1of the result is not satisfactory, the correct rate of other stages of sleep is satisfied.