Online signal to noise ratio improvement of ECG signal based on EEMD of synchronized ECG beats

Empirical Mode Decomposition (EMD) is a very powerful, signal dependent algorithm which decomposes signals as a set of Intrinsic Mode Functions (IMFs). The focus of this paper is on improving Signal to Noise Ratio (SNR) of noise contaminated Electrocardiogram (ECG) signal by applying a modified version of the Ensemble Empirical Mode Decomposition (EEMD) method. This method is utilized on synchronized sequential ECG beats of an ECG record. Since this method has a reasonable computational complexity and operates on the recorded signals, it can also be used in online applications. In this study, the achieved results with SNR of 12 for the EMD have been reported as 4.37×10^-4 in terms of Mean Square Error (MSE) and the MSE for the proposed EEMD for the same records have been reported as low as 1.08×10^-4. The experiments and results provided in this study have shown very promising performances compare to other methods such as simple EMD. In this paper, after confirming the fact that the intrinsic white noise is generally allocated to the first two IMFs of a contaminated ECG signal, it has been reported that the best results for the proposed EEMD method, in terms of MSE, have been achieved by removing the first two IMFs of the synchronized sequential beats of the input signals. Finally, the optimality and the efficiency of the proposed method have been evaluated in this paper by a comparison with two other methods, namely the EMD of the average signal and the simple averaging method.