DeepCEDNet: An Efficient Deep Convolutional Encoder-Decoder Networks for ECG Signal Enhancement

Abstract

Electrocardiogram (ECG) signal can be thought of as an effective indicator for detection of various arrhythmias. However, the acquired ECG data is always corrupted by amounts of noise, which have a great influence on the diagnosis of cardiovascular diseases. In this paper, an efficient deep convolutional encoder-decoder network framework is proposed to remove the noise from ECG signal, which is termed as 'DeepCEDNet'. This network is able to learn a sparse representation of data in the time-frequency domain via the high-order synchrosqueezing transform (FSSTH) and a nonlinear function that maps the noisy data into the clean one based on the distribution difference between signal and noise from the training set. Extensive experiments are conducted on ECG signals from the MIT-BIH Arrhythmia database and MIT-BIH Long-Term ECG database, and the added noise is obtained from the MIT-BIH Noise Stress Test database. The denoising performance is evaluated by means of signal to noise ratio (SNR), root mean squared error (RMSE) and percent root mean square difference (PRD). The results indicate that the proposed DeepCEDNet can obtain superior performance in both noise reduction and details preservation with higher SNR and lower RMSE and PRD compared to the traditional convolutional neural network (CNN) and the fully convolutional network-based denoising auto-encoder (FCN). We believe that the DeepCEDNet has a wide application prospect in the biomedical field.INDEX TERMS Electrocardiogram signal, noise reduction, deep neural network, sparse representation, time-frequency domain.

Channel-Wise Average Pooling and 1D Pixel-Shuffle Denoising Autoencoder for Electrode Motion Artifact Removal in ECG

“…However, this approach is limited because the noisy ECGs used are not realistic. In [19,21,22,25,27,28], the ratios of BW, EM, and MA artifacts in the noisy ECG were not mentioned, which indicates that the noisy ECGs may be different even if the same signal segments of noise and ECG are employed. Therefore, evaluating the denoising performance of each approach is not possible.…”

“…Several denoising approaches based on different techniques have been developed; examples include adaptive filtering [9][10][11], the wavelet method [12][13][14], empirical mode decomposition (EMD) [15][16][17][18], and denoising autoencoder (DAE) algorithms [19][20][21][22][23][24][25][26][27][28]. An adaptive filter updates its weight according to the error between the noisy ECG signal and the noise reference signal.…”

Lightweight Denoising Autoencoder Design for Noise Removal in Electrocardiography

“…Xiong et al first eliminated the noise in the ECG by DWT, and the remaining noise was further removed using a deep neural network (DNN)-DAE [24]. In [25], [26], a DAE architecture with a convolution neural network (CNN) was applied to remove noise in the ECG. The result reveals that the convolution layer extracts the clean feature effectivity more than the DNN and reduces the number of parameters.…”

ECG Denoising based on Transformation Matrix for Non-Decimated Wavelet Transform and Wavelet/Total Variation (WATV) Denoising

“…Many researchers intrested in denoising the ECG signal in the literature and proposed a great number of algorithms. Among those algorithms we can mention the Digital Filtering [6,7], Recursive Filtering [8,9], Adaptive Filtering [10,11], Wavelet Transform (WT) [12,13], Empirical Mode Decomposition (EMD) [14,15], Ensemble Empirical Mode Decomposition (EEMD) [16,17], Variational Mode Decomposition (VMD) [18], and deep learning-based technique [19,20] to name a few. In this paper, we propose a novel approach of Electrocardiogram (ECG) denoising which is based on Transformation Matrix for Non − Decimated Wavelet Transform (WT) [21] and Wavelet/Total Variation (WATV) Denoising [22].…”