Modeling of ECG Signal with Nonlinear Teager Energy Operator

Objective of this paper is to analyze noise free ECG signal and to discriminate the arrhythmia data from normal sinus rhythm data. ECG data is characterized by its nonlinear dynamic behavior, which shows significant changes between normal and arrhythmia data. Presence of artifacts like 60Hz power line interference (PLI), low frequency base line wander noise (2Hz) and muscle artifacts will not allow the analysis of ECG. In this paper two nonlinear modeling techniques, Multi Scale Principle Component Analysis (MSPCA) and Higher Order Spectral Analysis (HOSA) used for denoising of ECG data. In MSPCA principal components (PC) at multi scale are computed and those PC’s related to significant events are combined. MSPCA serves as powerful tool in denoising the ECG data from PLI noise. HOSA is valued for its characteristics of suppressing Gaussian noise for cumulants of higher order (3rdorder). This cumulant based method enhances the ECG data from artifacts like PLI, baseline wander , Gaussian noise and also improves the Signal to Noise Ratio. It is observed that HOSA is a simple nonlinear modeling technique that can be used to enhance the noisy ECG data. Teager Energy Operator (TEO), a nonlinear energy operator is applied on denoised ECG beat to find the energy generated by the source of signal rather than the energy of the signal itself. TEO of enhanced signal is valued for clear identification of arrhythmia data using parameters like Average nonlinear energy in time domain ANEt and average non linear energy in frequency domain ANEf . The performance measures are improvement in signal to noise ratio(SNR), Root Mean Square Error (RMSE), Root Mean Square Deviation (RMSD) [19]. TEO has successfully identified NSR data from arrhythmia data. The simulation work is carried out on a set of Normal sinus rhythm (NSR) and arrhythmia data taken from MIT-BIH database.