Abstract
Non-linear feature extraction is an effective method for processing EEG signal. Sample entropy, KC complexity and correlation dimension of EEG combined with the machine learning methods can get a good result in automatic sleep staging. However, due to the similar activity characteristics of EEG during light sleep N1 and REM, the discrimination of above methods in these two phases is reduced. That the brain is relatively active and sawtooth waves appear in REM will cause the amplitude's changes of EEG signal are dissimilar. This article proposes an improved feature extraction method based on the amplitude in order to improve the recognition effect of N1 and REM. Correlation dimension and sample entropy of EEG in N1 and REM are compared with the improved method so the effectiveness of the method introduced in this essay can be verified.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
