Correlation of Respiratory Signals and Electrocardiogram Signals via Empirical Mode Decomposition

Abstract

Recently Electrocardiogram (ECG) signals are being broadly used as an essential diagnosing tool in different clinical applications as they carry a reliable representation not only for cardiac activities, but also for other associated biological processes, like respiration. However, the process of recording and collecting them has usually suffered from the presence of some undesired noises, which in turn affects the reliability of such representations.Therefore, de-noising ECG signals became a hot research field for signal processing experts to ensure better and clear representation of the different cardiac activities. Given the nonlinear and non-stationary properties of ECGs, it is not a simple task to cancel the undesired noise terms without affecting the biological physics of them. In this study, we are interested in correlating the ECG signals with respiratory parameters, specifically the lung volume and lung pressure. We have focused on the concept of de-noising ECG signals by means of signal decomposition using an algorithm called the Empirical Mode Decomposition (EMD) where the original ECG signals are being decomposed into a set of intrinsic mode functions (IMF). Then, we have provided criteria based on which some of these IMFs have been adapted to reconstruct de-noised ECG version. Finally, we have utilized de-noised ECGs as well as IMFs for to study the correlation with lung volume and lung pressure. These correlation studies have showed some clear resemblance especially between the oscillations of ECGs and lung pressures.