An evaluation on the clinical outcome prediction of rotor detection in non-invasive phase maps.

Carlos Fambuena Santos1, Ismael Hernández-Romero2, Rubén Molero Alabau3, Andreu M. Climent3, Maria de la Salud Guillem Sánchez3
1UPV, 2University Rey Juan Carlos, 3Universitat Politècnica de València


Introduction. Electrocardiographic imaging (ECGI) and phase maps have been used in the past for rotor identification and ablation guidance in atrial fibrillation (AF). This study proposes a new rotor detection algorithm and evaluates its potential for clinical outcome prediction in patients undergoing pulmonary vein isolation (PVI). Materials and Methods. The performance of the new algorithm was evaluated using real AF episodes. Visually identified rotors were manually labelled using a custom application. Phase maps reconstructed from ECGI signals in 29 patients were calculated prior pulmonary vein isolation (PVI) and during adenosine injection. Different rotor metrics were calculated and compared in patients that recovered from PVI and patients that persisted in arrhythmia 6 months after ablation. These parameters were calculated applying also different rotor thresholds (0 and 1). Results. The mean precision and recall values of the algorithm were 0.82 and 0.75. A significantly higher concentration of phase singularities (PSs) in the PPVV in patients that recovered from PVI was found (0.16 vs 0.04, p < 0.01). In contrast, the number of rotors per second in the atria was higher in patients that did not recover from PVI (88.21 vs 68.55, p < 0.05). Conclusions. The obtained results suggest that rotor-related metrics obtained from phase maps and ECGI contain relevant information to predict clinical outcome in PVI patients.