04-05-2017, 01:08 PM
ECG thinning algorithm based noise reduction algorithms on empirical mode decomposition (EMD) and discrete wavelet transformation (DWT) domains. Unlike conventional EMD-based approaches that neglect a number of initial intrinsic functions (IMF) containing the QRS complex, as well as noise, it is proposed to construct windows in the EMD domain in order to reduce noise from IMF instead of completely discarding them, thus preserving the QRS complex and producing a relatively cleaner ECG signal. The signal thus obtained is transformed into the DWT domain, in which an adaptive soft threshold based noise reduction algorithm is employed considering the advantageous properties of the DWT in comparison to the EMD in the preservation of the energy in the presence of noise and in Reconstructing the original ECG With better resolution of time. Extensive simulations are performed using the MIT-BIH arrhythmia database and the performance of the proposed method is evaluated in terms of several standard metrics. The simulation results show that the proposed method is able to reduce the noise of noisy ECG signals more accurately and consistently compared to some of the more advanced methods.
The electrocardiogram (ECG) shows the electrical activity of the heart and is used by doctors to inspect the state of the heart. ECG analysis becomes difficult if the noise is embedded with the signal during acquisition. In this paper, a decomposition elimination technique is proposed for ECG signals based on Empirical Mode Decomposition (EMD). The ECG noisy signal is initially decomposed into a system of intrinsic mode functions (IMFs) using the EMD method. In the proposed technique, noise-dominated IMFs are automatically determined using Spectral Planarity (SF) measurement and then filtered using butterworth filters to eliminate noise. This method is evaluated in the ECG signals available in the MIT-BIH Arrhythmia database. The results of the experiment show that the proposed technique is performed with better Signal to Noise (SNR) and lower RMSE (Root Mean Square Error) than the commonly used Wavelet transformation technique.