Electrocardiogram Signals Error Correction Using Empirical Mode Decomposition Based Technique
Z.K Adeyemo, S.A Olayanju

Abstract
This paper presents the error correction of electrocardiogram signals which have been corrupted by Baseline wander and power line interference using empirical mode decomposition. The Electrocardiogram (ECG) is the electrical activity of the heart used by physicians to inspect the heart’s condition. The ECG recordings are often corrupted by artefacts. One of the dominant artifacts and interference present in ECG recordings is Baseline Wander (BW) that may be due to respiration or the motion of the patients or the instruments and electromagnetic Interference (EMI) from power line. Analysis of ECG becomes difficult if BW and power line interference are embedded with the signal during acquisition. Hence BW and power line interference need to be removed for better clinical evaluation. In this paper, a technique for removing BW and power line Interference in ECG is proposed. ECG signals available at MIT-BIH arrhythmia data base were used for the investigation. The original signal from MIT-BIH is first corrupted with both BW and Power line noise. The noisy ECG signal is initially decomposed into a set of Intrinsic Mode Functions (IMFs) using Empirical Mode Decomposition (EMD) method. The IMFs containing BW are filtered using a bank of lowpass filters, the noise rank was also determined and filtered using Infinite Impulse Response (IIR) notch filter. The clean ECG signal is derived from the combination of the processed IMFs. The simulations show that the proposed EMD-based method provides very good results for BW and Power line interference removal. Compared with the wavelet transform and EMD with spectral flatness, the method used gives better Signal Error Ratio (SER) Values. At SNR of 5dB, the wavelet transform method gives an SER of 8.8000; the EMD with spectral flatness gives SER of 9.9028: while the method used in this work gives SER of 10.1630

Full Text: PDF