Effect of Some Power Spectral Density Estimation Methods on Automatic Sleep Stage Scoring Using Artificial Neural Networks

Sleep staging has an important role in diagnosing sleep disorders. It is usually done by a sleep expert through examining sleep Electroencephalogram (EEG), Electrooculogram (EOG), Electromyogram (EMG) signals of the patients and determining the stages of sleep in different time sections named as epochs. Manual sleep staging is preferred among the sleep experts but because it is rather tiring and time consuming task, automatic sleep stage scoring systems get popularity. In this study, we obtained EEG, EMG and EOG signals of four healthy people at sleep laboratory of Meram Medicine Faculty of Necmettin Erbakan University to use them in sleep staging and extracted 20 different features by using some power spectral density estimation methods which are: Fast Fourier Transform (FFT), Welch and Autoregressive (AR). We evaluated the effects of these methods on sleep staging through using ANN classifier. Comparison between these methods was done on each individual whose data were utilized separately from others. According to the results, the maximum test classification accuracy was reported as 79.72% by using of FFT method for subject1. Also, mean of test classification accuracies for all of subjects were obtained as 74.14%, 71,58 and 70.34% with use of FFT, Welch and AR, respectively.