Online estimation of lower and upper bounds for heart sound boundaries in chest sound using Convex-hull algorithm

Area of Science:

• Biomedical Signal Processing
• Cardiovascular Acoustics
• Computational Medicine

Background:

• Accurate heart sound localization is crucial for effective heart sound cancellation algorithms.
• Precisely determining the onset and offset boundaries of heart sound segments presents a significant challenge.

Purpose of the Study:

• To develop a novel method for estimating heart sound segment boundaries.
• To establish reliable lower and upper bounds for onset and offset points.

Main Methods:

• Chest sounds are segmented into frames, and entropy and smoothed entropy features are extracted.
• The Convex-hull algorithm is applied to these features to estimate boundary envelopes.
• Heart sound boundaries are identified based on the difference between the envelope and entropy features exceeding a threshold.

Main Results:

• The proposed method demonstrates superior performance compared to a baseline method in heart sound localization.
• The novel approach achieved higher accuracy and a lower detection error rate.
• Smoothing entropy features significantly enhanced the performance of both proposed and baseline methods.

Conclusions:

• The developed method provides a robust approach for heart sound boundary estimation.
• The findings suggest that entropy-based features and the Convex-hull algorithm are effective for heart sound localization.
• Feature smoothing is a valuable technique for improving the accuracy of heart sound analysis.