Enhancement of the heart sound envelope using the logistic function amplitude moderation method

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Accurate segmentation of heart sounds (S1 and S2) is crucial for diagnosing cardiac conditions.
• Conventional methods struggle with noise interference and achieving uniform envelope peaks.

Purpose of the Study:

• To develop a novel method for extracting heart sound envelopes using the logistic function.
• To improve the discrimination between heart sounds and noise/silent intervals.
• To enhance heart sound segmentation accuracy, especially in low signal-to-noise ratio environments.

Main Methods:

• Utilized the logistic function's sigmoid characteristic to segregate S1 and S2 signal intensities.
• Developed the logistic function-based amplitude moderation (LFAM) envelogram method.
• Determined critical amplitude thresholds (x_uc, x_lc) via histogram analysis for noise and sound categorization.
• Evaluated performance on a noisy PCG dataset from PhysioNet/Computing in Cardiology Challenge 2016.
• Assessed the proposed envelope feature for heart sound segmentation using Hidden Semi-Markov Models (HSMM).

Main Results:

• The LFAM envelogram method demonstrated superior hill-valley discrimination of heart sounds from silent/noisy intervals.
• Enhanced heart sound envelope peaks showed improvement over conventional methods.
• Significant improvement in segmentation accuracy was observed, especially at low signal-to-noise ratios.
• Achieved a best average F1 score of 97.73% for heart sound segmentation.

Conclusions:

• The proposed logistic function-based method effectively extracts heart sound envelopes and improves segmentation accuracy.
• LFAM offers a robust approach for analyzing heart sounds in the presence of noise.
• This method holds promise for more accurate cardiac diagnosis through improved phonocardiogram analysis.