A Simple Automated VT/VF Detection Method for Resource-Constrained Life Threatening Arrhythmia Monitoring Devices

The accurate detection of ventricular fibrillation (VF) and ventricular tachycardia (VT) is crucial in an automatic external defibrillator (AED) and to prevent sudden cardiac arrests. This paper presents an automated and lightweight VT/VF detection method using a single time-domain feature, zero-crossing rate. The VT/VF detection method is tested with VT/VF and non-VT/VF segments of 5 -second duration from four databases, namely CU Tachyarrhythmia Database, MIT-BIH Malignant Ventricular Ectopy Database, MIT-BIH Arrhythmia Database, MIT-BIH Normal Sinus Rhythm Database, and St Petersburg INCART 12-lead Arrhythmia Database. The performance of the method is evaluated for electrocardiogram (ECG) segments of different morphologies across single and multiple leads. The sensitivity and specificity of $ {9 8. 6 0 \%}$ and $ {9 7. 2 2 \%}$, are obtained with 5 -second ECG segment sampled at 250 Hz. The method utilizes 0.429 MB of memory and achieves VT/VF detection in 0.79 seconds for a 5 -second ECG segment. Considering the resource limitations in AEDs and wearable VT/VF detectors, the lightweight method using infinite impulse response (IIR) filter-based noise suppression and zero-crossing rate feature provides a feasible solution for VT/VF detection during continuous monitoring.