Classification of electrocardiogram signals with waveform morphological analysis and support vector machines

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Electrocardiogram (ECG) signals are crucial for diagnosing cardiac diseases.
• Accurate arrhythmia classification is vital for automated medical diagnosis.
• Existing methods may require further optimization for speed and precision.

Purpose of the Study:

• To develop a novel, accurate, and efficient ECG signal classification method.
• To combine diverse signal features for improved diagnostic performance.
• To reduce the time required for ECG signal classification.

Main Methods:

• ECG signal denoising followed by wavelet packet decomposition.
• Extraction of frequency-domain features (singular value, max value, std dev) from coefficients.
• Extraction of time-domain features (RR intervals) using the slope threshold method.
• Fusion of time and frequency features into a comprehensive feature space.
• Classification using Support Vector Machine (SVM) with grid search and morphological analysis.

Main Results:

• The proposed method achieved a classification accuracy of 96.67% on arrhythmia databases.
• Combination of waveform morphology and frequency domain analysis enhanced accuracy.
• The approach minimized signal classification time.

Conclusions:

• The novel multi-feature classification method demonstrates high accuracy and efficiency for ECG-based arrhythmia diagnosis.
• Combining time and frequency domain features provides a robust approach.
• This method holds promise for improving automated cardiac disease diagnosis.