The Computer-assisted Automatic Sleep Scoring System Based On EEG

As sleep involves in physiology, psychology, neurobiology, medical science,computer science and rehabilitation medicine, recently many countries have established independent "Sleep Research Society" organizations to strengthen and enhance the study of sleep. The sleep staging comes first on the study and analysis of sleep EEG signal. Past sleep stage work has long been artificial completed by experts, but the training rules between experts are often difficult to obtain consistence. Automatic sleep stages gradually becomes major concern in late 80 s, but the current conventional sleep staging system recognition rate is not high. So we consider designing a automatic scoring system of sleep EEG signal with high efficiency and accuracy.This paper firstly discusses the current situation of domestic and foreign study of sleep staging and the importance of sleep EEG signal used in sleep staging. Secondly it applies the approximate entropy, wavelet transform, detrended fluctuation analysis and fuzzy neural system in sleep staging. The work involved is below:(1) We introduce the working principle of wavelet denoising and the pretreatment of sleep eeg. DB4 wavelet is used in 5 level decomposition for signal denoising.(2) According to the characteristics of sleep EEG, we extract approximate entropy, the relative wavelet energy and DFA index of sleep EEG. By the comparison of their features in different period of sleep, we find their different regular changes will help sleep index score.(2) Introducing the fuzzy neural network based on the ANFIS model, we train three different types of feature parameters to explore scoring system with highest efficiency and accuracy. Ultimately we determine that the sleep scoring system of REW and scaling exponent with ANFIS has high recognition, low computational cost and fast computing speed which can also be realized real-time in the hardware.