Wavelet Based Sleep EEG Detection Using Fuzzy Logic

The Sleep stage classification has been accomplished using fuzzy inference system, where the prerecorded data of sleep EEG has been processed with the help of wavelet transform. The investigation on sleep stage detection reveals the quantitative presence of three different stages of sleep i.e. Awake, SWS (slow wave sleep) and REM (rapid eye movement). The proposed work approaches to correctly identify the three classes of sleep EEG using fuzzy classification method based on fuzzy rule base. The 3-channel data is preprocessed by wavelet transform via signal processing tools and further processed to identify the stages of sleep EEG. The extracted features from the processed data are EEG sub-band frequencies, standard deviation measures for EMG and EOG and variance measures for EMG and EOG. These features are required to make the fuzzy rules for FIS (Fuzzy inference system) and further used to identify the sleep stages correctly. Performance analysis of the proposed fuzzy model was accurately evaluated in terms of fuzzy variables and the result shows that the proposed approach is able to classify the EEG signals with the average accuracy of 93% in which SWS stage was best detected among other stages of sleep EEG.