Noise reduction for heart sounds using a modified minimum-mean squared error estimator with ECG gating.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Heart sound recordings are often contaminated by various noise sources, hindering accurate analysis.
• Existing noise reduction methods may not sufficiently preserve signal integrity or effectively remove diverse noise types.

Purpose of the Study:

• To develop and evaluate an advanced single-channel noise reduction method for heart sound recordings.
• To improve the signal-to-noise ratio (SNR) and reduce distortion in heart sound signals.

Main Methods:

• Utilized spectral domain minimum-mean squared error (MMSE) estimation with a decision-directed approach for noise spectrum estimation.
• Incorporated modifications including soft thresholding, forward-backward filtering, and a second-pass iterative scheme.
• Integrated electrocardiogram (ECG) analysis for gating to guide noise spectral estimation.

Main Results:

• The proposed algorithm demonstrated superior performance compared to existing methods in terms of SNR gain.
• Qualitative evaluations indicated a significant reduction in noise and signal distortion.
• Improvements were observed in the automatic detection of abnormalities in heart sounds when using the algorithm as a front-end.

Conclusions:

• The developed noise reduction technique effectively enhances the quality of heart sound recordings.
• This method offers a promising solution for improving the accuracy of automated heart sound analysis systems.
• The integration of ECG gating further refines the noise reduction process for better diagnostic potential.