Improved ventricular fibrillation/tachycardia detection using NEWFM for automated external defibrillators

Abstract

Ventricular fibrillation (VF) and ventricular tachycardia (VT) are life-threatening signals. Automated external defibrillators can decrease the fatality rate if the VF/VT detection is stable and quick. This thesis proposes improved VF/VT detection. For our experiments, we use the complete Creighton University Ventricular Tachyarrhythmia Database. Samples are analyzed under the same conditions in intervals of 7 s. Based on this data, we propose a time-delay transform. Then, we extract six shockable features, three known and three new, which are used to construct our Neural Network with Weight Fuzzy Membership Functions model (NEWFM). The result is better than the phase space reconstruction algorithm. © 2015 SERSC.