Electrocardiogram morphological arrhythmia classification using fuzzy entropy-based feature selection and optimal...

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Cardiac arrhythmias contribute significantly to global mortality.
• Accurate detection of arrhythmias requires continuous ECG monitoring and advanced analysis.
• Computer-assisted algorithms are crucial for interpreting complex ECG data.

Purpose of the Study:

• To develop and validate a novel morphological arrhythmia classification algorithm for ECG signals.
• To identify and rank significant features for improved classification accuracy.
• To compare the performance of different machine learning classifiers for ECG beat categorization.

Main Methods:

• A novel feature set of 25 attributes was extracted from ECG beats.
• Fuzzy Entropy-based Feature Selection (FEBFS) was employed to rank and select features.
• Support Vector Machine with Radial Basis Function (SVM-RBF) and Weighted K-Nearest Neighbor (WKNN) classifiers were utilized.
• Performance was evaluated using 10-fold cross-validation on the MIT-BIH Arrhythmia Database.

Main Results:

• The WKNN classifier, with K=3 and cityblock distance, achieved the highest accuracy.
• Achieved Average Sensitivity = 94.89%, Positive Predictivity = 97.13%, Specificity = 99.72%, F1 Score = 95.95%, and Overall Accuracy = 99.15%.
• The proposed algorithm demonstrated superior performance compared to existing state-of-the-art methods.

Conclusions:

• The developed algorithm, utilizing a unique feature set and FEBFS, is efficient and reliable for morphological arrhythmia classification.
• The findings highlight the potential of advanced signal processing and machine learning for improving cardiac arrhythmia detection.
• This work offers a robust solution for beat-by-beat ECG analysis in clinical settings.