Signal identification and noise suppression in multi-channel ECG and MCGby Independent Component Analysis (ICA)

Independent Component Analysis (ICA) is a method to decompose a multi-channel signal into statistically independent components. An overview over a second order blind identification algorithm is given and implications of its constraints for the application on multi-channel ECG and magnetocardiogram (MCG) are discussed. We tested parameters for that ICA algorithm that led to a good separation of low frequency noise and power line interference from the heart signal. The usefulness of ICA to separate different physiological signals is shown in examples from a fetal ECG recording and in ECGs/MCGs from atrial flutter patients. While the separation of noise and physiological signal yields stable results, the identification of distinct physiological components of the heart signal by ICA can only be achieved in special cases and needs further investigations.