Eliminating cardiac electrical artifacts from cardiac autonomic nervous signals using a combination of empirical mode decomposition and independent component analysis

Cardiac autonomic nervous (CAN) signals in ambulatory dogs can nowadays be measured by an implantable radio transmitter system. CAN signals are known to be related to heart failure. However, they are critically contaminated by cardiac electrical activities (CEA) which confound data analysis. We propose a method of analysis which combines empirical mode decomposition (EMD) and independent component analysis (ICA). This method composed of two steps: First, the EMD method decomposed a single channel recording into multichannel data, then we applied the ICA to these multichannel data. Using an ambulatory dog's CAN signal data from Seoul National University Hospital, we compared our approach with a commonly used high pass filter (HPF) method for various amplitudes of simulated CAN signals. Root-mean-squared errors between simulated CAN signals and CAN signals with CEA artifact were calculated for assessing the noise cancellation effect. Moreover, we observed changes in spectral content via power spectral density. Finally, we applied the proposed method to real data. Our method could not only extract and remove CEA artifact in CAN signals, but also preserved the spectral content of CAN signals.