Denoising and Extraction of Electrocardiogram Signal Using EPMD

Abstract

Signal processing a major tool used for ECG analysis and interpretation in today's life. ECG signal is used to identify the different types of diseases. In ECG signal processing is used to improve the measurement accuracy and reproducibility comparatively. Separating respiration signals from ECG is one way of obtaining knowledge related to respiration especially when specialized equipments are not used to monitor the respiration continuously. There are chances of noises added to ECG signal when it is transferred via wireless medium. Some of these noises are baseline wander, power – line interference, etc. These types of noises corrupt the ECG signal resulting in a way unable to diagnose the disease. However, removing the noise can only give the exact ECG signal. Even with this we cannot identify the diseases. In order to overcome this drawback we are using Ensemble Pragmatic Mode Decomposition technique to remove these types of noises with single channel ECG based on higher order statistics. The proposed approach is altered from the existing NLMT algorithm in few aspects: the transform domain collaborative filtering and the block-based processing. The waveform of respiratory signal is reconstructed from the NLWT by processing single-channel ECG. Two techniques for the decomposition of the ECG signal into suitable bases of functions are proposed Hilbert-Huang Transform (HHT) Analysis and the Ensemble Pragmatic Mode Decomposition (EPMD) to achieve the goals. The frequency information evolving with time scales and time locations provides the performance of HHT and Ensemble Pragmatic Mode Decomposition by an analysis of Intrinsic Mode Function (IMF). This technique is used to overcome the drawbacks of wavelet approach and to extract the respiratory signal separately to easily identify the different types of diseases.