Background:
Pulmonary vein isolation (PVI) is a cornerstone for atrial fibrillation (AF) ablation. AF recurrence rate after ablation remains high and repeat procedures still a challenge, regardless of the ablation strategy.

Spatio-temporal electrograms dispersion are a visually recognizable electrical footprint maintaining AF. However, electrogram guided ablation could be subjective and not reproducible between operators.

Objective:
We sought to investigate the benefit of new software technology based on artificial intelligence (AI), to guide selective ablation of electrograms exhibiting a spatiotemporal dispersion in repeat ablation for persistent AF.

Methods:
We present the preliminary results of a single center experience of repeat ablation in patients with persistent and long-standing persistent AF. Nine patients were included in the analysis. All patients had at least one previous pulmonary vein isolation (PVI) ± substrate ablation. A high-density mapping of both atria using a 20-pole catheter and a 3D mapping system was performed. Regions displaying electrogram dispersion (ED) during AF as detected by the AI software were tagged. Repeat PVI was performed if necessary. The further step was to target ED areas, with connection to the nearest non conduction area if located <1cm. Atrial tachycardias (AT) arising during the procedure were then mapped and ablated. The ablation endpoint was termination into sinus rhythm (SR).

Results:
The ablation strategy led to AF termination via AT in 4 patients and to a significant AF cycle length prolongation in 5 patients. The pulmonary veins were encountered isolated in 88% of patients (n= 8 patients). ED areas were located mostly on the anterior LA (n=8 patients), followed by the posterior wall (n= 7 patients). The right atrium was considered as an ablation target in 88% of all patients (n= 8 patients) . An additional individualized approach including anterior line, anteroseptal line and/ or roof line targeting Ablation at the previous tagged ED points was performed in 7 patients. The mean procedure and ablation time was 210 ± 45 min and 40 ± 27min, respectively. No complications occurred.

Conclusions:
In patients undergoing repeat ablation procedure(s) for persistent AF, targeting spatio-temporal electrogram dispersion guided by an artificial intelligence software had a significant acute impact on AF and might lead to promising long-term results.