Analysis Of Sleep Electroencephalogram Signal Based On Wavelet Transformation

Sleep is one of the most important physiological phenomena for vertebrate. People use electroencephalogram (EEG) as a very powerful tool to analyze sleep. We can get plenty of physiology and pathology information from EEG, so it can be used for clinical diagnosis of sleep-related disease and for the study of our brains. This paper introduces the principle of wavelet transform and wavelet packet decomposition, and uses them for raw EEG de-noising and basic rhythm extraction. Then we conduct sleep EEG wavelet energy analysis of different sleep stages on the basis.EEG signals are produced by the electrophysiological response of the brain nervous system. The signals are weak, non-stationary, low signal-to-noise ratio and easy to be effected by EOG, ECG, EMG and white noise. This paper uses adaptive threshold de-noising based on the wavelet packet to pre-process the raw EEG signals and achieves good results. Then we extract the basic rhythm using the wavelet transform.The research on sleep staging is mainly based on the study of EEG rhythm α and δ waves and their energy ratio. The wavelet energy has significant differences in the various stages of sleep and presents a change trend of regularity. It has the maximum value in the lucid stage, becomes smaller in the NREM stage Ⅰ and Ⅱ, reach the minimum value in the stage Ⅲ and Ⅳ,and rises again to the value close to stage Ⅰ when the REM stage starts. It has well-defined boundaries between different stages, except for stage Ⅰ and REM stage. The comparison result with artificial sleep staging shows that the energy ratio of EEG rhythm α and δ waves in different sleep stages can be used for sleep staging. It is a new characteristic variable and provides a new approach for sleep staging.