DGM-TOP: automatic identification of the critical boundaries in atrial tachycardia

Robin Van Den Abeele^1*Sander Hendrickx¹Niels Carlier¹Eike Moritz Wülfers¹Arthur Santos Bezerra¹Bjorn Verstraeten¹Sebastiaan Lootens¹Karel Desplenter¹Arstanbek Okenov¹Timur Nezlobinsky¹Annika Haas²Armin Luik²Sebastien Knecht³Mattias Duytschaever³Nele Vandersickel¹

• ¹Ghent University, Ghent, Belgium
• ²Städtisches Klinikum Karlsruhe, Karlsruhe, Baden-Württemberg, Germany
• ³AZ Sint-Jan Brugge-Oostende AV, Brugge, West Flanders, Belgium

In the latest research on topology in cardiac arrhythmia, it was demonstrated through a fun-damental mathematical principle called the index theorem that reentry based atrial tachycar-dias (AT) are maintained by pairs of counter-rotating waves that are either complete or near-complete rotations. Each wave is centered around a different anatomical object that exhibitsa non-zero index/topological charge, called a critical boundary. Interconnecting both criticalboundaries with an ablation line terminates the tachycardia.This research focuses on the specific algorithms for calculating the index/topological charge ofeach anatomical boundary, called DGM-TOP. This is achieved by using information from theelectroanatomical map of the patient, more specifically, the sequential activation time of nodesaround those boundaries. With this method, pairs of critical boundaries were consistently de-tected in 100% of the 578 in-silico and 100% of the 24 clinical ATs. Adhering to the previouslydescribed index theorem. Additionally, ablation results in both datasets show that terminationof AT is only possible by interconnecting both critical boundaries. This outcome highlights theimportance of detecting the critical boundaries before deciding on the correct ablation line, asany ablation line that does not connect both critical boundaries is unable to terminate the AT.Moreover, in the case of incorrect ablation, the BCL-algorithm was proposed to estimate theincrease in tachycardia cycle length. However, only moderate correlation (r2 = 0.62) is ob-served for simulations, indicating a refinement of this BCL-algorithm is necessary in additionto a larger clinical dataset.