Empirical mode decomposition based ECG features in classifying and tracking ventricular arrhythmias.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Ventricular arrhythmias (VA), including ventricular tachycardia (VT) and ventricular fibrillation (VF), are critical conditions affecting heart function.
• VF is particularly lethal, with its progression and response to therapy significantly impacting cardiac resuscitation success.
• Time-critical quantification and rapid feedback are essential for managing VA, especially VF.

Purpose of the Study:

• To develop computationally efficient, data-driven techniques for classifying and tracking VA over time.
• To provide tools for characterizing VA dynamics in 'in-hospital' settings for long-term therapy planning.
• To enable near real-time feedback for 'out-of-hospital' VA detection and progression monitoring to guide immediate therapeutic decisions.

Main Methods:

• Utilized Empirical Mode Decomposition (EMD) for analyzing electrocardiogram (ECG) data of VA.
• Developed distinct approaches for 'in-hospital' (detailed characterization) and 'out-of-hospital' (real-time feedback) scenarios.
• Classified VA types: VT vs. VF, and sub-classified VF into organized VF (OVF) vs. disorganized VF (DVF).

Main Results:

• Achieved high classification accuracies in 'in-hospital' analysis: 96.7% for VT vs. VF and 87.2% for OVF vs. DVF.
• Demonstrated strong potential for monitoring VA progression over time.
• Reported average accuracies of 71% for VT/VF and 73% for OVF/DVF in near real-time 'out-of-hospital' analysis.

Conclusions:

• The developed EMD-based techniques are computationally efficient and effective for classifying and tracking VA.
• These methods offer significant potential to assist clinicians in therapy planning and medical personnel in real-time treatment adjustments.
• The study highlights the utility of data-driven signal processing for improving outcomes in life-threatening cardiac arrhythmias.