Feasibility of using patient-specific models and the "minimum cut" algorithm to predict optimal ablation targets for left atrial flutter

Abstract

Background: Left atrial flutter (LAFL) occurs in patients after atrial fibrillation ablation. Identification of optimal ablation targets to terminate LAFL remains challenging.

Objective: The purpose of this study was to use patient-specific models to simulate LAFL and predict optimal ablation targets using a novel approach based on flow network theory.

Methods: Late gadolinium-enhanced cardiac magnetic resonance scans from 10 patients with LAFL were used to construct atrial models incorporating fibrosis by investigators blinded to procedural findings. Rapid pacing was applied in silico to induce LAFL. In each LAFL, we represented reentrant wave propagation as an electric flow network and identified the "minimum cut" (MC), which was the smallest amount of tissue that separated the flow into 2 discontinuous components. In silico ablation was applied at MCs, and targets were compared to those that terminated LAFL during catheter ablation.

Results: Patient-specific atrial models were successfully generated from patient scans. LAFL was induced in 7 of 10 models. Ablation of MCs terminated LAFL in 4 models and produced new, slower LAFL morphologies in the other 3. For the latter cases, flow analysis was repeated to identify MCs of emergent LAFLs. Ablation of these MCs terminated emergent LAFLs. The MC-based ablation lesions in simulations were similar in length and location to ablation targets that terminated LAFL during catheter ablation for these 7 patients.

Conclusion: Personalized atrial simulations can predict ablation targets for LAFL. These simulations provide a powerful tool for planning ablation procedures and may reduce procedural times and complications.

Keywords: Ablation target; Fibrosis; Left atrial flutter; Network theory; Patient-specific atrial model.

Figure 1

Model generation and identification of minimum cuts in reentrant flow propagation. I: Pipeline to construct magnetic resonance image–based models of fibrotic human atria. A: Representative late gadolinium-enhanced cardiac magnetic resonance (LGE-CMR) slice of human atria. B: Atrial tissue segmentation into fibrotic (green) and nonfibrotic (gray) regions. C: Three-dimensional reconstruction of atrial geometry. D: Fiber orientation in atrial model. E: Atrial action potentials in fibrotic (green) and nonfibrotic (gray) tissue. II: Schematic for identifying minimum cuts in reentrant flow propagation, represented as a flow network. The minimum cut of the flow network is indicated with the dashed line.

Figure 2

Three-dimensional (3D) patient-specific atrial models. Personalized 3D models of all 10 patient atria reconstructed from late gadolinium-enhanced cardiac magnetic resonance scans.

Figure 3

Examples of left atrial flutter sustained by peripulmonary vein reentry. Transmembrane maps showing reentry around the left inferior pulmonary vein in model 1 (A) and figure-of-eight reentry in model 2 (B). The counterclockwise circuit in the figure-of-eight reentry is around the left inferior pulmonary vein and the clockwise circuit is in the posterior left atrium. LAA = left atrial appendage; LIPV = left inferior pulmonary vein; LSPV = left superior pulmonary vein; MV = mitral valve; RIPV = right inferior pulmonary vein; RSPV = right superior pulmonary vein.

Figure 4

Examples of left atrial flutter sustained by perimitral reentry. Transmembrane voltage maps showing figure-of-eight reentry in model 3 (A) and model 4 (B). In both cases, the clockwise circuit in the figure-of-eight reentry is around the mitral valve and the counterclockwise circuit is in the posterior left atrium. LAA = left atrial appendage; LIPV = left inferior pulmonary vein; LSPV = left superior pulmonary vein; MV = mitral valve; RIPV = right inferior pulmonary vein; RSPV = right superior pulmonary vein.

Figure 5

Example of left atrial flutter sustained by reentry around fibrotic tissue in the posterior left atrium. Transmembrane voltage maps showing reentry around fibrotic tissue in the posterior left atrium for model 5. Fibrotic tissue is highlighted in green. LAA = left atrial appendage; LIPV = left inferior pulmonary vein; LSPV = left superior pulmonary vein; MV = mitral valve; RIPV = right inferior pulmonary vein; RSPV = right superior pulmonary vein.

Figure 6

Location of minimum cuts in reentrant flow propagation and termination of left atrial flutter (LAFL) by ablation. Location of minimum cuts (cyan) of reentrant flow propagation for model 2 (A), model 3 (B), and model 4 (C) overlaid on activation maps of LAFL episodes. Transmembrane voltage maps showing LAFL termination in model 2 (D), model 3 (E), and model 4 (F) after in silico ablation was applied (red). LAA = left atrial appendage; LIPV = left inferior pulmonary vein; LSPV = left superior pulmonary vein; MV = mitral valve; RIPV = right inferior pulmonary vein; RSPV = right superior pulmonary vein.

Figure 7

Emergence of postablation left atrial flutter (LAFL). Location of minimum cuts (cyan) of reentrant flow propagation for model 1 (A) and model 5 (B) overlaid on activation maps of LAFL episodes. Transmembrane voltage maps showing emergence of new LAFL in model 1 (C) and model 5 (D) after in silico ablation was applied (red). LAA = left atrial appendage; LIPV = left inferior pulmonary vein; LSPV = left superior pulmonary vein; MV = mitral valve; RIPV = right inferior pulmonary vein; RSPV = right superior pulmonary vein.

Figure 8

Comparison of in silico and clinical ablation targets (those outside of pulmonary vein isolation lines). Ablated minimum cut in models 1 (A, i), 2 (B, i), 3 (C, i), 4 (D, i), and 5 (E, i). Ablated tissue to terminate left atrial flutter in clinical electrophysiologic study for patients 1 (A, ii), 2 (B, ii), 3 (C, ii), 4 (D, ii), and 5 (E, ii).