The Analysis And Classify Of Sleep Stage Using Deep Learning Network From Single-Channel Eeg Signal

Electroencephalogram (EEG)-based sleep stage analysis is helpful for diagnosis of sleep disorder. However, the accuracy of previous EEG-based method is still unsatisfactory. In order to improve the classification performance, we proposed an EEG-based automatic sleep stage classification method, which combined convolutional neural network (CNN) and time-frequency decomposition. The time-frequency image (TFI) of EEG signals is obtained by using the smoothed short-time Fourier transform. The features derived from the TFI have been used as an input feature of a CNN for sleep stage classification. The proposed method achieves the best accuracy of 88.83%. The experimental results demonstrate that deep learning method provides better classification performance compared to other methods.