An enhanced EMD algorithm for ECG signal processing

Abstract

The Empirical Mode Decomposition (EMD) is becoming increasingly popular for the multi-scale analysis of signals. However, the data-driven and adaptive nature of the EMD raises concerns regarding the uniqueness of the decomposition as well as the extend to which oscillatory modes can be mixed across different IMFs. This paper proposes a solution to this problem for the analysis of ECG signals. The bivariate extension of the decomposition (BEMD) is used as the basis of an analysis in which a synthetic ECG signal of idealized waveform guides the decomposition of an input ECG segment. Essentially, this work provides the necessary ground for the deployment of signal processing algorithms on the ECG signal using a more robust EMD analysis.