HYBRID WAVELET-KERNEL MACHINES IN ENDOCARDIAL SIGNAL ANALYSIS

摘要

Rate based arrhythmia recognition algorithms in implantable car-dioverter-defibrillators are of limited reliability in some clinical situations. Here the inclusion of morphological features of endocardial electrograms can improve the performance. In this study, we present a coupled signal-adapted wavelet-support vector machine (SVM) arrhythmia detection scheme. Within the scope of an electrophysiological examination, data segments were recorded during normal sinus rhythm (NSR) and ventricular tachycardia (VT). Consecutive beats were selected as morphological activation patterns of NSR and VT. These patterns were represented by their multilevel concentrations. For this, a signal-adapted and highly efficient lattice structure based wavelet decomposition technique was employed which maximizes the class separability and takes the final classification of NSR and VT by hard margin SVMs with radial compactly supported kernels into account. In an automated analysis of an independent test-set, our hybrid scheme outperformed other methods and classified all patterns correctly without overlap.