Time-frequency analysis of a single ECG: to discriminate between ventricular tachycardia and ventricular fibrillation

Accurate discrimination of ventricular tachycardia (VT) and ventricular fibrillation (VF) still remains a challenge in automatic external defibrillators. Over the last decades, a lot of algorithms in the time and frequency domain have been proposed. Although good results have been obtained in both domains, the time-frequency distribution (TFD) represent an ideal tool to improve them. The aim of this study is to examine if Choi-Williams and Cone Kernel TFD allow one to characterize SR, VT and VF; then estimate adequate parameters to achieve an algorithm able to discriminate between SR, VT and VF. This study has shown that discrimination of VT from VF by analyzing a single ECG lead using Choi-Williams and Cone Kernel is feasible. Both TFD achieved similar results over 90%. The results are better than algorithms based on the time or frequency domain, although not so much as was envisaged