| With the continuously rapid development of social economy, the work pressure is increasing.More and more people pay attention to the quality of sleep and sleep disorders. Therefore,classifying the sleep EEG correctly by studying sleep EEG has a very important significance.This paper is based on above-mentioned goal. It introduces two methods to study the sleepEEG and sleep stage classification. They are detrended cross-correlation analysis (DCCA) andmultifractal detrended cross-correlation analysis (MF-DCCA).Firstly, this thesis mainly researched detrended cross-correlation analysis. The DCCA wasdesigned to investigate power-law cross correlations between different simultaneously recordedtime series in the presence of nonstationarity. In this paper, the method of detrendedcross-correlation analysis was used to analyze sleep stage classification. The sleepelectroencephalograph signals were extracted from the MIT-BIH Polysomnographic Databaserandomly. The results showed that the average DCCA exponent of the awake period is smaller thanthat of the first stage of non-rapid eye movement sleeps. The method was of great significance todistinguish the quality of sleep, or diagnose and treat of sleep disorders.Secondly, the thesis introduced another method for researching sleep stage classification. Thatwas multifractal detrended cross-correlation analysis. On the basis of theoretical analysis, themultifractality was combined with detrended cross-correlation to describe the cross-correlationbetween the time series. According to the average multifractal spectrum width, sleep stages couldbe distinguished. The experiments showed that the average multifractal spectrum width of soberperiod is less than that of the sleep stageâ… . Sleep stages could be distinguished well by the method.The method for studying the sleep electroencephalography has a great significance to determine thequality of sleep or diagnosis and treatment of sleep disorders. |
PDF Full Text Request