Synchrosqueezed Transform based Click Event Segmentation in Phonocardiogram of Mitral Valve Prolapse

Area of Science:

• Biomedical Engineering
• Cardiology
• Machine Learning in Healthcare

Background:

• Phonocardiogram (PCG) signals from mitral valve prolapse (MVP) patients exhibit characteristic transient audio events like systolic clicks and murmurs.
• Traditional auscultation for detecting these clues is error-prone, potentially leading to missed diagnoses and unnecessary expensive tests like echocardiography.

Purpose of the Study:

• To develop an automated machine learning technique for accurate detection of MVP from PCG signals.
• To assist clinicians in identifying critical auscultation clues, thereby improving diagnostic accuracy and efficiency.

Main Methods:

• A novel segmentation technique was employed utilizing Fourier synchrosqueezed transform (FSST) features.
• A long short-term memory (LSTM) neural network was integrated with FSST features for signal analysis and classification.

Main Results:

• The proposed FSST-LSTM method achieved a high accuracy of 99.8% on the MVP dataset.
• The technique demonstrated significant potential for reliable clinical diagnosis of MVP.

Conclusions:

• The developed machine learning approach offers a promising, accurate, and automated solution for MVP detection using PCG signals.
• This method can serve as a valuable tool to support clinicians, reducing diagnostic errors and improving patient care pathways.