Breakthrough Wave Detection in a 3D Computer Model of Atrial Endo-Epicardial Dissociation 

Éric IRAKOZE and Vincent Jacquemet
University of Montreal


Abstract

INTRODUCTION: Breakthrough (BT) activation patterns have been observed in experimental mapping of atrial fibrillation. Endo-epicardial (endo-epi) dissociation is hypothesized to cause such pattern by creating intramural 3D pathways. We aim to reproduce in a computer model this mechanism which might be misinterpreted as ectopic activity.

METHODS: A 3D cubic-mesh atrial model (280 µm discretization) with locally controllable endo-epi dissociation was created. Numerical methods were developed to introduce epi-endo uncoupling in a regular-grid finite difference model while appropriately handling boundary conditions and anisotropy despite staircase representation and wall thinness. Propagation was simulated using Courtemanche-based monodomain equations. Fibrillation was initiated in the isolated epicardial layer. Then, epi-endo discrete connections were introduced as N disks of 4-mm diameter randomly distributed over the atrial surface (16 models with up N = 284 connections, as well as a uniformly coupled case). In each model, 5 s of fibrillation was simulated. BT were automatically identified by tracking waves from membrane potential maps of the epicardial surface, and manually validated.

RESULTS: BT patterns were observed near most endo-epi connections and only there. There was no BT in the uniformly coupled model. Higher BT rates (> 50/s) were observed in models with intermediate number of endo-epi connections (32 ≤ N ≤ 128), while BT rate was smaller (< 50/s) in highly coupled (N > 128) or sparsely coupled (N < 32) models, in agreement with previous studies in 2D bilayer models. Instances of stable repetitive BT occurrences were identified, consisting of up to 8 consecutive BT with a cycle length of about 130 ms. These activation patterns looked similar to a focal source.

CONCLUSION: Epi-endo dissociation combined with discrete coupling can be implemented in 3D cubic mesh models to create transmural reentries and BT patterns. Such models may be used to validate mapping signal analysis software.