A Noise Elimination Method for ECG Signals

To solve the problem of lacking adaptability and difficulties in selecting wavelet base when using the Wavelet-ICA to cancellation the noise in ECG, a new method called E-ICA is proposed which is combined with the Empirical Mode Decomposition and Independent Component Analysis. The advantages of EMD and ICA are exploited, and the two layers reconstruction, that is inverse ICA and inverse EMD was introduced to get the clean ECG. Using the MIT-BIH database to simulate, the results show that E-ICA has a better capacity to remove the noise in ECG, the correlation coefficient of clean ECG and original ECG is 0.96.