Patient-Specific Identification of Atrial Flutter Vulnerability–A Computational Approach to Reveal Latent Reentry Pathways

Figures

• 

  Figure 1. Overview of the algorithm to compute vulnerability maps considering all possible AFlut loops (left) and transfer the AFlut scenarios to dynamic simulations (right).

• 

  Figure 2. Example of an AFlut vulnerability map for the right atrium (anterior aspect in A, posterior aspect in B). The vulnerable pathways are marked in yellow on the substrate of the right atrium (brown); the blood pool is indicated in gray. Several pathways run from the septal side of the tricuspid valve to the right atrial appendage.

• 

  Figure 3. Fitted exponential restitution curves of ERP (A,B) and CV (C,D), as well as the WL as the product of the former measures (E,F) for different anatomical regions in the atria (A,C,E) and different substrates (B,D,F). In (B,D,F), the dashed lines represent the respective substrates under the influence of 2.3 μM amiodarone whereas the dotted lines represent the influence of 0.21 μM dronedarone. Exponential curves according to Equation (3) were used to fit the output of monodomain tissue strand simulations. The CV for the homgeneous control model in (D) was reduced compared to the RA/LA tissue in the heterogeneous setup in (C) to obtain a similar total atrial activation time. In (A), BB and PM curves are covered by the RA/LA curve. Coefficients are listed in Supplementary Table 1. RA, right atrium; LA, left atrium; CT, crista terminalis; PM, pectinate muscles; BB, Bachmann's bundle; II, inferior isthmus; PV, pulmonary veins; RAA, right atrial appendage; LAA, left atrial appendage; TVR, tricuspid valve ring; MVR, mitral valve ring.

• 

  Figure 4. Loop candidate finding. Activation time resulting from a stimulus at the junction of the superior vena cava with the right atrial appendage (yellow star) ranging from early (blue) to late (red). The direction of activation is indicated by arrows (A). Points of wavefront collisions were detected and are indicated by red dots in (B). (C) shows the loops composed of the two traces leading from the collision site to the stimulus site (yellow star). The yellow loops fulfill the WL condition Equation (5) whereas the light gray loops do not and were thus not considered for further steps.

• 

  Figure 5. Evolution of a geometric snake covering the LA. Initially, the snake (gray) covered the loop found by the collision tracing algorithm (blue band) corresponding to a RTT of 390 ms (A). By iteratively constricting the snake (B–F), the shortest RTT of 304 ms considering heterogeneous CV and anisotropy was found. In this way, the influence of the particular choice of the stimulus site (yellow star in A) was reduced.

• 

  Figure 6. Influence of heterogeneity and anisotropic conduction on vulnerability maps of the RA. Vulnerable pathways are marked in yellow on the RA myocardium (brown); the blood pool is indicated in gray. While CV was tuned such that the latest sinus rhythm activation coincided in all models, both anisotropy and heterogeneity lead to a higher number of vulnerable pathways. (A) isotropic homogeneous; (B) anisotropic homogeneous; (C) isotropic heterogeneous; (D) anisotropic heterogeneous.

• 

  Figure 7. Sensitivity of vulnerable pathway coverage to changes in homogeneous, anisotropic CV (A) and the stimulus point density (B). The CV was altered in a homogeneous, anisotropic setup causing different degrees of RA coverage by vulnerable pathways for a fixed stimulus density of 1 mm. In (B), the distance between stimulus points was varied for a fixed CV of 425 mm/s.

• 

  Figure 8. Vulnerability maps of the RA for combinations of different substrates and pharmacological agents. Besides a control substrate representing healthy myocytes, a cAF remodeled substrate (neglecting changes of cell-to-cell coupling), and the two hERG mutations N588K and L532P were evaluated. Standard concentrations of the antiarrhythmic agents amiodarone (2.3 μM) and dronedarone (0.21 μM) were administered in the center and right columns, respectively. Vulnerable pathways are marked in yellow on the RA myocardium (brown); the blood pool is indicated in gray. Supplementary Figure 4 shows the results for a matched total sinus rhythm activation time in all setups.

• 

  Figure 9. Vulnerability maps of the RA for different substrate modifications. Compared, to (A), a zone of slow conduction was introduced in (B,C) (light brown area enclosed by dashed circle). A circular ablation lesion (black line) was introduced in (D). In (E,F), a gap in the ablation lesion with varying extent was modeled on the central posterior wall.

• 

  Figure 10. Application of the proposed simulation pipeline on a clinical example. The inital vulnerability map (A) shows numerous pathways along which AFlut can be sustained. By closing the gap in the previous line on the left part of the anterior wall (green arrow), the number of pathways could be reduced but a significant amount remained (B). The extrapolated phase map (C) for one of the pathways identified in (B) (gray line in D) was used to intialize a dynamic fast marching simulation (D) exhibiting stable reentry. Additional ablation to the (RSPV) (green arrow) eliminated all vulnerable pathways (E).