Complexity Comparison of Empirical Mode Decomposition and Wavelet Decomposition Methods in the Detection of Ventricular Late Potential

Abstract

Ventricular Late Potentials (VLPs) mark the electrical instability of myocardial tissues of the heart. It has been observed in medical researches that there is a direct link between sudden cardiac death due to arrhythmia and the presence of VLPs. Early detection of CVDs can be made possible by the detection of VLPs. The wavelet-based decomposition is the most widely used method in literature however due to the multiple stages involved in the wavelet-based decomposition the computational complexity of the system is high. This paper proposes VLP detection method using the Empirical Mode Decomposition (EMD) which is simple and more efficient. The ECG signal is initially filtered and the consecutive individual beats in the ECG are averaged to obtain the Signal Averaged ECG (SAECG). The EMD is applied to the obtained SAECG which decomposes the signal into corresponding Intrinsic Mode Functions (IMFs) from which the presence of VLPs can be detected. The proposed method captures even the lowest intensity deviation present in a signal. In addition to this the Wavelet decomposition is found to be two times more complex compared to the EMD based method with respect to the number of samples given.