Improving Discrimination of Ventricular Tachycardia and Ventricular Fibrillation Using Classifier Ensembles and High Dimensional Representations

Research output: Contribution to journalArticlepeer-review

179 Downloads (Pure)

Abstract

Differentiating between ventricular tachycar- dia and ventricular fibrillation is a difficult problem, and highly relevant for clinical practice and translational research. While having low false arrhythmia alarm rates, previous approaches were found to be inadequate for discriminating between these two categories. For improv- ing their discrimination, we introduce the use of high- dimensional feature vectors, in particular, magnitude spec- tra, and classifier ensembles that take into account local information from the electrocardiogram signals. In order to deal with the increased false arrhythmia alarm rate that results from this approach, a hierarchical classification is proposed, which significantly improves the classification sensitivities of ventricular tachycardia and ventricular fibrillation, while at the same time achieving a modest decrease in the false alarm rate.
Original languageEnglish
Pages (from-to)1-10
Number of pages10
JournalIEEE Transactions on Biomedical Engineering
Publication statusAccepted/In press - 2014

Cite this