Background: We have previously demonstrated, in a pilot study, the utility of personalized computational atrial modeling to guide ablation of persistent AF (psAF) in patients with fibrosis detected by LGE-MRI. However, the methodology was heavily operator-dependent.
Objective: To develop a comprehensive and minimally operator-dependent computational platform for reproducible personalized psAF ablation target prediction in patients with LGE-MRI fibrosis, so that our FDA-approved randomized clinical trial on the novel ablation approach can be undertaken.
Methods: We designed a computational platform that streamlines the end-to-end modeling workflow for predicting clinical ablation targets, with minimal human interaction. A series of automatic software interactions were implemented to integrate the platform’s modules, requiring only the atrial segmentation as the input. The platform consists of 3 main modules 1) 3D model reconstruction, 2) Personalized simulation and 3) Personalized ablation strategy plan compatible with clinical system (Fig. 1).
Results: This platform was successfully tested by multiple operators with differing expertise on 10 psAF patients. No modifications nor human interactions to the platform were required during testing. The same ablation target was predicted for each patient when processed by multiple operators, highlighting the robustness and reproducibility of the platform. This optimized pipeline reduced the model reconstruction time by 75% (16 vs 4 hours).
Conclusion: We developed a robust, reproducible, automatic platform for personalized psAF ablation target prediction, enabling our FDA-approved randomized clinical trial.