Heart murmur classification with feature selection.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Heart sounds provide vital information for assessing cardiac function.
• Abnormal heart sounds, such as murmurs, indicate potential cardiovascular issues like valve dysfunction.
• Accurate heart murmur classification is crucial for diagnosing heart valve disorders.

Purpose of the Study:

• To develop and evaluate a novel method for heart murmur classification.
• To improve the accuracy and efficiency of diagnosing heart valve disorders.
• To reduce the complexity of feature sets used in heart murmur analysis.

Main Methods:

• A new set of 17 features was extracted from heart sound signals in time, frequency, and state-space domains.
• Feature selection was performed using the Floating Sequential Forward Selection (SFFS) method, reducing the feature set to 10.
• A nonlinear classifier was employed for murmur classification.
• Performance was validated against established methods on a common database.

Main Results:

• The proposed method successfully reduced the feature set from 17 to 10 features.
• The classification achieved using the reduced feature set demonstrated slightly improved results compared to existing state-of-the-art methods.
• The method proved effective in distinguishing between different types of heart murmurs.

Conclusions:

• The proposed feature extraction and selection method offers a more efficient approach to heart murmur classification.
• This technique holds potential for improved non-invasive diagnosis of cardiovascular heart diseases.
• The findings suggest that a smaller, carefully selected feature set can yield superior classification performance.