Leveraging conditional diffusion and pruning for optimal ECG signal denoising.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Electrocardiograms (ECGs) are crucial for cardiac diagnosis but are susceptible to noise and baseline wander.
• Existing denoising methods face limitations in preserving signal integrity or computational efficiency.
• Advanced techniques often require significant computational resources, hindering clinical application.

Purpose of the Study:

• To develop an efficient and accurate ECG denoising method.
• To address the limitations of traditional and advanced ECG noise reduction techniques.
• To improve the diagnostic reliability of ECG signals in real-world clinical settings.

Main Methods:

• Implementation of an Improved Denoising Diffusion Probabilistic Model (IDPM).
• Integration of a conditional framework and Quality Assignment Pruning technique.
• Leveraging diffusion models for complex dataset processing and noise reduction.

Main Results:

• The proposed IDPM method significantly outperforms state-of-the-art and traditional ECG denoising models.
• Achieved precise noise reduction and baseline correction while preserving essential ECG features.
• Demonstrated superior performance even under extreme noise conditions.

Conclusions:

• The IDPM offers a robust and efficient solution for ECG denoising.
• Reduced computational requirements make it suitable for low-resource clinical devices.
• This approach has potential for broader applications in biomedical signal processing.