Sleep EEG Staging Based On Hilbert-Huang Transform And Sample Entropy

Sleep staging has clinical and practical significance that it has an important role in the evaluation of sleep quality and adjuvant therapy of sleep disorders. The traditional manual sleep stage has its limitations:low efficiency, time consuming and human cost, so the research of automatic sleep staging has important significance. EEG is one of the most important physiological signals for sleep, by a process of sleep EEG, extract the characteristic parameters which can reflect the different sleep stages, and with the help of classifier for sleep staging. Sleep EEG is a complex, time-varying, nonlinear and non-stationary signal, this paper extracted characteristics with methods of nonlinear dynamics of sample entropy and time-frequency analysis method, Hilbert-Huang transform method. Calculate the EEG energy ratio of each sleep stage through the HHT and then combine the sample entropy characteristic. All the sleep characteristics as the input of the classifier, and using the Libsvm toolbox developed by Dr. Lin Zhiren of National Taiwan University for classification of sleep stage. The experimental data used in this paper is come from the Sleep-EDF database of MIT-BIT PhysioBank, and selected two EEG of 10 subjects for sleep staging. This paper mainly divided sleep into the awake period, NREM 2 period, NREM 3 period (deep sleep), NREM 1/REM (rapid eye movement).The experimental results shows that, use sample entropy and Hilbert-Huang transform can effectively obtain the sleep EEG characteristics. A certain regularity exists between different sleep stages of sample entropy, in non rapid eye movement (NREM) stage, along with in-depth, sample entropy decreases, and reached the minimum at NREM 3,4. The marginal spectrum of EEG signals by using Hilbert-Huang transform has some differences in different stage of sleep, the EEG energy ratio can characterize different sleep stages. But the effect by only using the sample entropy for sleep stage is so-so, and by using Hilber-Huang transform for the sleep feature extraction, the sleep stage effect is better. Sleep feature extraction by combining the sample entropy and Hilbert-Huang transform, the effect of sleep stage is improve, better than using only one method, the overall accuracy rate reach 89.9%. Thus, the sleep staging is valid through the method combined with sample entropy and Hilbert-Huang transform method as feature extraction, at the same time also prove the feasibility of EEG for sleep stages.