Computational Modeling Guided Ablation for Atrial Flutteris

Atypical left atrial flutter (ALAF) is a relatively stable arrhythmia commonly seen in patients after they have undergone ablation for atrial fibrillation. ALAF can be difficult to rate control and frequently requires another ablation. These require extensive mapping to delineate the circuit; there may even be undetectable latent circuits that arise after ablating the dominant one. Mapping is further complicated by very low voltages due to extensive existing scar, making correct local activation time determination very difficult. At times the only solution can be treatment by atrio-ventricular (AV) node ablation and pacemaker implant. ALAF is set up around scar regions in the left atrium. We propose a novel, multidisciplinary approach to managing ALAF in these difficult cases using personalized computational modeling of atrial flutter to guide ablation. Within our atrial arrhythmia group, we have pioneered the use of MRI for measuring atrial structure and scar, developed a wide range of experimental approaches to extend mechanisms of electrophysiology, and applied state-of-the- art scientific computing, all based on highly integrated local resources. We will apply this multidisciplinary expertise to improve the understanding and outcomes of post-ablation ALAF. To achieve this goal, we will (1) use the scar information to make personalized computational models of atrial flutter to identify circuits and provide guidance to ablation, (2) improve MRI segmentation and thresholding accuracy to determine post- ablation scar in patients and (3) use the post ablation MRI and outcomes to determine the mechanistic underpinnings of procedural success and failure in patients who undergo computational model guided ablation for ALAF at the Salt Lake VA. Left atrial late gadolinium enhancement techniques will be improved to acquire isotropic images. Machine learning will be used in our extensive database of thousands of left atrial MRIs at the Salt Lake VA and the Univ of Utah and hundreds of ALAF cases to develop a faster and more accurate segmentation and thresholding technique. Computational model will be used to carry out virtual electrophysiology studies to induce flutter with the goal of identifying all the flutter circuits. Further simulations will be used to test ablation strategies to eliminate all possible left atrial flutters. Our preliminary studies support these goals by achieving successful prediction of arrhythmias from computational models, at times better than the most advanced mapping techniques. The result of this project will be a generalized computational model of ALAF, which can be personalized with left atrial geometry, scar pattern, and mapping results. Such a model will enable virtual electrophysiology studies leading to the prediction of ablation points to terminate all possible ALAF pathways. If successful this will lead to improving outcomes, reducing the need for repeat ablations that have significant cost and risks associated with them and possibly even avoid doing AV node ablations.