Analysis of phonocardiogram signals through proactive denoising using novel self-discriminant learner

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Phonocardiogram (PCG) signals contain vital cardiac health information but are susceptible to noise and motion artifacts.
• Accurate analysis of PCG signals is crucial for detecting abnormal cardiac conditions.
• Existing denoising methods often struggle with the complexity of PCG signal noise.

Purpose of the Study:

• To develop a robust and automated denoising technique for Phonocardiogram (PCG) signals.
• To enhance the accuracy of clinical event detection and physiological abnormality identification from PCG data.
• To introduce a novel self-discriminant learner for distinguishing clean from noisy PCG signals without manual intervention.

Main Methods:

• Development of Proclean, a denoising technique utilizing pattern recognition and statistical learning.
• Implementation of a novel self-discriminant learner for feature extraction to differentiate clean and noisy PCG signals.
• Validation using publicly available MIT-Physionet datasets.

Main Results:

• Proclean achieves over 85% accuracy in detecting noisy PCG signals.
• The proposed denoising mechanism improves physiological abnormality detection by more than 20%.
• The self-discriminant learner effectively distinguishes clean from noisy PCG signals without human oversight.

Conclusions:

• The Proclean technique offers a robust solution for denoising PCG signals, crucial for accurate cardiac health assessment.
• Automated noise detection and elimination in PCG analysis lead to significantly improved clinical insights.
• This approach enhances the reliability of PCG-based diagnostic tools for medical investigations.