Detection of ventricular arrhythmia using hybrid time-frequency-based features and deep neural network

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Sudden cardiac death (SCD) is a primary cause of mortality in cardiac patients, often resulting from ventricular tachyarrhythmia (VTA), encompassing ventricular tachycardia (VT) and ventricular fibrillation (VF).
• Timely prediction of VTA and prompt use of automated external defibrillators (AEDs) are critical for improving survival rates.
• Current diagnostic methods face challenges in rapid and accurate VTA detection.

Purpose of the Study:

• To develop and evaluate a deep neural network (DNN) based classification scheme for distinguishing between VF and VT conditions.
• To enhance the prediction accuracy and speed of VTA detection using hybrid time-frequency features.
• To compare the performance of the proposed DNN model against standard machine learning algorithms for VTA classification.

Main Methods:

• Utilized two public ECG databases (CUDB and VFDB) for training, testing, and validation.
• Employed hybrid time-frequency-based features extracted after signal decomposition using wavelet transform, empirical mode decomposition (EMD), and variable mode decomposition (VMD).
• Developed a deep neural network (DNN) classifier to analyze the extracted features from 5-second ECG signal windows.

Main Results:

• The DNN classifier achieved high performance metrics: 99.2% accuracy (Acc), 98.8% sensitivity (Se), and 99.3% specificity (Sp).
• The proposed DNN approach demonstrated superior performance compared to other standard machine learning algorithms.
• The algorithm accurately detects VTA conditions, showing potential to reduce misinterpretations by human experts.

Conclusions:

• The developed DNN model offers an accurate and efficient method for classifying VF/VT conditions.
• This approach can significantly improve the efficiency of cardiac diagnosis through ECG signal analysis.
• The findings suggest a promising tool for reducing mortality associated with sudden cardiac death by enabling faster and more accurate VTA detection.