An ECG Signal Noise Reduction Method Based on Improved Residual Dense Network

Abstract

An ECG signal denoising method based on an improved residual dense network. The residual dense network has the ability to reuse features, which reduces the computational cost while achieving ECG signal denoising. In the process of applying the residual dense network, there is no need to artificially set parameters based on experience, avoiding empirical errors, and improving the generalization ability of the model. In the improved residual dense network, the input of each improved residual block is fused with the output of all the previous improved residual blocks; by removing the noise of the ECG signal through this network, the output of all the previous improved residual blocks can be obtained. Output, enhanced feature propagation; as the network deepens, there will be no problems such as gradient messages and gradient explosions. Considering the local and global features of the signal at the same time, it can not only capture the local features of the signal and preserve useful medical features, but also capture the global features of the signal and stabilize the training process.

Description

A Denoising Method of ECG Signal Based on Improved Residual Dense Network technical field The present invention relates to the technical field of ECG signal noise reduction, be specifically related to a kind of heart based on improved residual dense network Electrical signal noise reduction method. Background technique [0002] The electrocardiogram is the main basis for the diagnosis of cardiovascular disease, and is an important auxiliary means for cardiovascular doctors to check. ECG signals contain important medical information such as physiology and pathology, and reflect the physiological health of various parts of the heart to a certain extent. important biomedical signals. However, due to the weak and random characteristics of ECG signals, and often in clinical applications Accompanied by a large amount of noise, it is easy to cause ECG waveform distortion, which will affect the identification of each band of the signal, and even affect the medical treatment. ...

AI Technical Summary

Problems solved by technology

However, ECG signals often contain a variety of noises, and the noise types cannot be determined, and the effect of a single denoising method is not ideal

In the traditional noise reduction method, due to the aliasing of the noise signal and part of the ECG signal waveform, it is easy to cause signal waveform distortion and lose a lot of useful information

Some deep learning denoising methods, such as autoencoders, generative confrontation network methods, etc., have problems such as gradient disappearance or gradient explosion, so it is difficult to train deeper networks, which affects subsequent research on ECG signals

Images