Objective To analyze the characteristics of FastICA based on negentropy and kurtosis, and the application of ECG denoising based on the two kinds of FastICA in order to study the characteristics of ICA in separation of ECG and noise. Methods Synthetic and real ECG contaminated by noise are applied to test the two FastICA. Synthetic ECG consists of random noise and single ECG from MIH-BIT to test algorithms’speed and accuracy. Real ECG is from DaISy database to analyze the two ICA methods’effects in the real environment. Results Through synthetic and real ECG tests, the two kinds of FastICA are proved to be effective in ECG denoising. And results show that FastICA based on kurtosis has higher speed and FastICA based on negentropy has higher accuracy. Conclusions The two kinds of FastICA can be effectively applied in ECG denoising with good characteristics.
|