A simplicity-based fuzzy clustering approach for detection and extraction of murmurs from the phonocardiogram

Area of Science:

• Cardiology
• Biomedical Signal Processing
• Medical Diagnostics

Background:

• Cardiac abnormalities often present as murmurs on phonocardiograms (PCG).
• Accurate localization of murmurs within the cardiac cycle is crucial for diagnosing cardiac dysfunction.
• Existing methods struggle with murmur localization due to spectral similarity, time overlap, and variations in amplitude and spectral characteristics.

Purpose of the Study:

• To develop a robust algorithm for accurately locating cardiac murmurs within phonocardiograms (PCG).
• To overcome limitations of existing methods by focusing on murmur visual simplicity rather than absolute amplitude and frequency.
• To enable better diagnosis of cardiac dysfunction through improved murmur isolation.

Main Methods:

• A novel method is proposed that leverages the visual simplicity of murmurs for localization.
• Fuzzy sets are employed to cluster simplicity values and determine murmur duration.
• The algorithm's performance is evaluated based on accuracy, sensitivity, and specificity in detecting systolic murmurs.

Main Results:

• The proposed algorithm demonstrates an overall accuracy of 80% in detecting systolic murmurs.
• The sensitivity for locating systolic murmurs is 73%, with a specificity of 100%.
• The method successfully isolates murmurs, allowing for further feature extraction for clinical diagnosis.

Conclusions:

• The visual simplicity-based approach offers a more reliable method for localizing cardiac murmurs in PCG signals.
• This technique improves upon existing algorithms by being less dependent on absolute murmur characteristics.
• The enhanced localization of murmurs facilitates more accurate diagnosis of cardiac abnormalities.