TEO separated AM-FM components for identification of apnea EEG signals

Sleep apnea event is occurred due to hindrance in respiration, which is most commonly observed in children and adults. It is noticed that, if this event sustained for long time it will cause several brain and heart disorders. Diagnoses of such disorders require identification of apnea event. Recently, electroencephalogram (EEG) signals show proficiency for assessing the sleep quality and detection of apnea event. In this work, amplitude modulated (AM) and frequency modulated (FM) components based features extracted from EEG signals, are using for identification of sleep apnea event. Teager energy operator (TEO) is uses for separation of AM-FM components but it requires band limited signals. The nature of EEG is highly non-stationary, so an adaptive empirical mode decomposition technique is applied, which convert the non-stationary EEG signals into band limited intrinsic mode functions (IMF). TEO separates each IMF into AM-FM components. The extracted features from separated components are applied as input to least square support vector machine (LS-SVM) classifier and obtained better performance parameters for identification of apnea event compared to existing methods.