Filtering the magnetohydrodynamic effect from 12-lead ECG signals using Independent Component Analysis

When recording an electrocardiogram (ECG) under the presence of the static magnetic feld of a magnetic resonance (MR) scanner, the ECG signal is disturbing by the magnetohydrodynamic (MHD) effect. Due to the MHD effect, the diagnostic information contained in the ECG signal is not accessible. This article copes with this problem by applying different Independent Component Analysis (ICA) algorithms to ECG signals corrupted by a simulated and real MHD effect. The applied ICA methods lead to significant errors in the resulting, filtered ECGs. Especially for the ECG signals with real MHD effect, a separation of the ECG and MHD components was not possible.