Ecg-based arrhythmia classification using discrete wavelet transform and attention-enhanced CNN-BiGRU model.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiology

Background:

• Electrocardiogram (ECG) signal noise impedes accurate arrhythmia classification.
• Early detection of cardiovascular diseases through ECG analysis is critical.

Purpose of the Study:

• To develop a robust ECG-based arrhythmia classification system.
• To enhance classification performance by addressing signal noise and class imbalance.

Main Methods:

• Discrete Wavelet Transform (DWT) for ECG signal denoising.
• Borderline-SMOTE for addressing class imbalance.
• Attention-Enhanced Convolutional Neural Network-Bidirectional Gated Recurrent Unit (CNN-BiGRU) for classification.

Main Results:

• Achieved 99.22% accuracy in classifying five arrhythmia categories on the MIT-BIH database.
• The proposed method outperformed existing arrhythmia detection techniques.
• Successfully preserved essential ECG morphological features during denoising.

Conclusions:

• The DWT and Attention-Enhanced CNN-BiGRU model offer an effective solution for automatic arrhythmia detection.
• This approach holds promise for improving clinical diagnosis of cardiovascular conditions.
• The method demonstrates high discriminative capability for identifying arrhythmias.