A study of sleep staging based on a sample entropy analysis of electroencephalogram

In this paper we report a detection method for different sleep stages and it is based on a single-channel electroencephalogram (EEG) system. The system is simple and can be easily setup in homes to perform sleep EEG recording, overnight sleep EEG automatic staging, and sleep quality evaluation. EEG data of 14 sleeping subjects were recorded through the entire night. All subjects were within the age group of 20-30 years and having no significant sleep disorders. To analyze the EEG data, it is segmented into equal time intervals. This is followed by calculation of Sample Entropy (SampEn) for each section, and the SampEn's statistical characteristics, such as the median, upper quartile, lower quartile and inter-quartile range. The sleep data were divided into training group (7 cases) and test group (7 cases). Sleep stages' quantitative ranges of training group referring to ZEO results were extracted and the quantization range used to sleep staging EEG data. Both the training group and test group results were close to ZEO results. It suggested that the statistical characteristics of Sample Entropy could be used as a criterion for sleep staging and evaluation.