Novel DERMA Fusion Technique for ECG Heartbeat Classification

Area of Science:

• Cardiology and Biomedical Engineering
• Signal Processing and Machine Learning

Background:

• Electrocardiogram (ECG) analysis relies on identifying waveform characteristics (P, QRS, T) and time intervals to detect cardiac abnormalities.
• Accurate classification of various heartbeats, such as premature ventricular contraction (PVC), left bundle branch block (LBBB), right bundle branch block (RBBB), PACE, and atrial premature contraction (APC), is crucial for diagnosing heart conditions.

Purpose of the Study:

• To develop and evaluate a novel fusion technique for enhanced identification and classification of abnormal and normal morphological events in ECG signals.
• To accurately classify five distinct types of heartbeats using advanced signal processing and machine learning models.

Main Methods:

• Feature extraction was performed on ECG signals to identify key components like P, QRS complex, and T waves.
• A fusion technique combining dual event-related moving average (DERMA) for peak analysis and fractional Fourier-transform (FrlFT) for time-frequency analysis was proposed.
• Two supervised learning models, Support Vector Machine (SVM) and K-Nearest Neighbor (KNN), were trained for heartbeat classification.
• Experiments utilized two datasets: MIT-BIH Arrhythmia and the Shaoxing and Ningbo People's Hospital (SPNH) database.

Main Results:

• The proposed DERMA and FrlFT fusion technique demonstrated high efficacy in identifying abnormal and normal ECG events.
• The automated model achieved exceptional performance in classifying five types of heartbeats.
• The system achieved an accuracy of 99.99%, sensitivity of 99.96%, and specificity of 99.9%.

Conclusions:

• The fusion of DERMA and FrlFT algorithms provides a robust and highly accurate method for ECG signal analysis and cardiac condition classification.
• The developed automated model, trained with supervised learning techniques, significantly advances the potential for real-time cardiac abnormality detection.
• The study highlights the effectiveness of combining signal processing and machine learning for precise diagnosis of various arrhythmias.