Sleep Apnea Frame Detection Based on Empirical Mode Decomposition of Delta Wave extracted from Wavelet of EEG Signals.

Abstract

In this paper, we have proposed an apnea framedetection method based on the Empirical Mode Decomposi-tion(EMD) of wavelet reconstructed delta wave of EEG signal.The method begins with wavelet transforming an EEG frameand reconstructing the low frequency delta wave from theapproximate coefficients. EMD is carried on the reconstructeddelta wave to generate intrinsic mode functions(IMF). Meanrate of variation and variance in the first five IMFs of thereconstructed delta wave are extracted as features from eachframe. Finally SVM classifier is used to test the performance ofthe proposed method. From MIT-BIH sleep apnea database, theproposed method is tested with 13 overnight polysomnographic(PSG) records. The proposed method is applied on each patientand overall patients. We found accuracy, sensitivity and specificityrate of 80.43%, 85.59% and 77.87% respectively on overallpatients. In conclusion, our proposed method is an efficient method for detecting apnea and non-apnea frames when onlyEEG signal is available and can be a great tool for PSG SleepApnea diagnosis.