Detection of the Third Heart Sound Based on Nonlinear Signal Decomposition and Time-Frequency Localization.

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• The third heart sound (S3) is a critical indicator of heart failure.
• Detecting S3 is challenging due to its low energy and frequency, often leading to misidentification with abnormal S2 sounds.
• Existing literature lacks methods to address the misinterpretation of low-energy S3.

Purpose of the Study:

• To develop a precise method for detecting the third heart sound (S3).
• To overcome challenges associated with low energy, frequency, and potential misidentification of S3.
• To improve the accuracy of heart failure diagnosis through enhanced S3 detection.

Main Methods:

• Nonlinear single decomposition using the Hilbert vibration decomposition method to preserve phase information.
• Time-frequency localization of decomposed subcomponents via smoothed pseudo Wigner-Ville distribution and reassignment.
• Distinguishing S3 based on positional information and time delays relative to the second heart sound (S2).

Main Results:

• The proposed method successfully detects S3 even with normalized temporal energy above 0.16 and frequencies above 34 Hz.
• Achieved a high accuracy rate of 93.9% in S3 detection across 82 cardiac cycles from diverse databases.
• Demonstrated superior performance compared to existing methods in identifying low-energy S3 sounds.

Conclusions:

• The developed method offers a robust and accurate approach for detecting low-energy S3 sounds.
• This advancement has significant implications for improving the early diagnosis of heart failure.
• The technique effectively addresses the limitations of previous S3 detection methods.