Detection of Shockable Ventricular Arrhythmia using Variational Mode Decomposition.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Ventricular tachycardia (VT) and ventricular fibrillation (VF) are critical cardiac arrhythmias requiring prompt detection for defibrillation.
• Accurate detection of VT/VF is essential for automated external defibrillator (AED) and implantable cardioverter defibrillator (ICD) therapies.
• Existing methods for arrhythmia detection require improvement in accuracy and specificity.

Purpose of the Study:

• To propose and evaluate a new method for detecting and classifying shockable (VT/VF) and non-shockable ventricular arrhythmias from ECG signals.
• To utilize variational mode decomposition (VMD) for ECG signal analysis.
• To assess the efficacy of machine learning classifiers in identifying cardiac arrhythmias.

Main Methods:

• ECG signals were decomposed into modes using variational mode decomposition (VMD).
• Energy, Renyi entropy, and permutation entropy of the first three modes were extracted as diagnostic features.
• Mutual information was used for optimal feature selection.
• A random forest (RF) classifier was employed for arrhythmia classification.

Main Results:

• The proposed method achieved high performance metrics: 97.23% accuracy, 96.54% sensitivity, and 97.97% specificity.
• The feature subset selected via mutual information scoring demonstrated strong diagnostic capability.
• The RF classifier effectively distinguished between shockable and non-shockable ventricular arrhythmias.

Conclusions:

• The developed method offers a robust and accurate approach for detecting shockable ventricular arrhythmias from ECG.
• The combination of VMD, mutual information feature selection, and RF classification shows significant promise for clinical application in defibrillator therapy.
• This technique provides a valuable advancement in the real-time monitoring and treatment of life-threatening cardiac arrhythmias.