A denoising method of ECG signal based on variational autoencoder and masked convolution

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Wearable electrocardiogram (ECG) devices enable continuous cardiovascular monitoring but are prone to noise interference, compromising diagnostic accuracy.
• Effective noise reduction is crucial for reliable ECG signal interpretation in wearable health technology.

Purpose of the Study:

• To propose a novel noise reduction model for ECG signals.
• To enhance the diagnostic correctness of wearable ECG monitoring devices by mitigating noise interference.

Main Methods:

• A noise reduction model integrating variational autoencoder (VAE) and masked convolution is developed.
• Variational Bayesian inference is employed within the VAE to capture global ECG signal features.
• Masked convolution modules are utilized to extract and integrate local ECG signal features, enhancing overall performance.

Main Results:

• The proposed model significantly improves signal-to-noise ratio (SNR) and reduces root mean square error (RMSE) compared to existing methods.
• Experimental results on the MIT-BIH arrhythmia database demonstrate superior noise reduction capabilities.
• The model effectively reduces signal distortion while enhancing noise reduction performance.

Conclusions:

• The novel VAE and masked convolution-based model offers a significant advancement in ECG signal noise reduction.
• This approach holds promise for improving the reliability and accuracy of wearable cardiovascular disease monitoring.
• The method provides a robust solution for enhancing the quality of ECG data in real-world applications.