Traversed Graph Representation for Sparse Encoding of Macro-Reentrant Tachycardia

Abstract

Macro-reentrant atrial and ventricular tachycardias originate from additional circuits in which the activation of the cardiac chambers follows a high-frequency rotating pattern. The macro-reentrant circuit can be interrupted by targeted radiofrequency energy delivery with a linear lesion transecting the pathway. The choice of the optimal ablation site is determined by the operator’s experience, thus limiting the procedure success, increasing its duration and also unnecessarily extending the ablated tissue area in the case of incorrect ablation target estimation. In this paper, an algorithm for automatic intraoperative detection of the tachycardia reentry path is proposed by modelling the propagation as a graph traverse problem. Moreover, the optimal ablation point where the path should be transected is computed. Finally, the proposed method is applied to sparse electroanatomical data to demonstrate its use when undersampled mapping occurs. Thirteen electroanatomical maps of right ventricle and right and left atrium tachycardias from patients treated for congenital heart disease were analysed retrospectively in this study, with prediction accuracy tested against the recorded ablation sites and arrhythmia termination points.