Detecting ventricular fibrillation by fast algorithm of dynamic sample entropy

A key component in automatic external defibrillator (AED) is to discriminate ventricular fibrillation (VF) from non VF by means of appropriate detection algorithms, which should have a high detection quality, be easily implementable, and work in real time. In this study, we presented a novel and computationally simple algorithm for VF detection using sample entropy (SampEn) method. To evaluate the accuracy of the proposed algorithm, we analyzed the complete data sets from Creighton University Ventricular Tachyarrhythmia database (CUDB). We compared the sensitivity, specificity, positive prediction, accuracy and the area under its receiver operating characteristic curve (ROC) of the new algorithm with several earlier VF detection algorithms. The new algorithm proves to be well suited for short data sets analysis, which can reach an elegant balance between detection time and accuracy.