ECGDeDRDNet: A Deep-Learning-Based Method for Electrocardiogram Noise Removal Using a Double Recurrent Dense Network

It is challenging to find an effective Electrocardiogram (ECG) noise reduction method in remote ECG monitoring systems. Most existing methods do not consider various noise conditions present in real-world scenarios. In remote monitoring, the acquisition environment is more complex, with a wide variety of noise that significantly reduces the clinical diagnostic accuracy of ECG. To address this, we propose ECGDeDRDNet, a deep learning-based ECG denoising framework leveraging a Double Recurrent Dense Network architecture. Unlike traditional approaches, we introduce a double recurrent scheme to enhance information reuse from both ECG waveforms and the estimated clean image. For ECG waveform processing, our base model employs LSTM layers cascaded with DenseNet blocks. The estimated clean ECG image, obtained by subtracting predicted noise components from the noisy input, is iteratively fed back into the model. This dual recurrent architecture enables comprehensive utilization of both temporal waveform features and spatial image details, leading to more effective noise suppression. Experimental results on the MIT-BIH dataset demonstrate that our method achieves superior performance compared to classical ECG denoising techniques in terms of SNR and RMSE. We also develop an image-based denoising version by converting ECG signals into 2D images. Results show competitive performance against conventional image denoising methods in terms of PSNR and SSIM. However, further validation on additional datasets is needed to establish generalizability.