590Development of a patient-specific computational fluid dynamics model of the LA in AF for stroke risk assessment

Abstract

Purpose: To develop a patient-specific computational fluid-dynamics (CFD) model of the left atrium (LA) to analyze hemodynamic mechanisms and their interplay in atrial fibrillation (AF) conditions, aimed at improving stroke risk stratification and therapy optimization. Method: Dynamic CT imaging of a patient in persistent AF was used to derive the 3D LA anatomical model applying a specifically designed segmentation algorithm based on image histogram and optimized with morphological operators and curvature motion. A registration step was applied to obtain the patient-specific LA motion field throughout the cardiac cycle from CT data. To perform CFD simulation in the LA, the arbitrary Lagrangian Eulerian formulation of the Navier-Stokes equations was applied in both sinus rhythm (SR) and AF conditions. SR simulation was performed considering the LA motion field previously obtained; AF simulation was performed applying a noisy motion field. In both cases, boundary conditions were set imposing a pseudoparabolic flow profile at each pulmonary vein (PV). Then, considering a suitable mitral valve flow-rate (the same for the two conditions apart from the absence of the A wave for the AF simulation) and the LA volume change throughout the cardiac cycle, the flow-rate at each PV was assigned by applying the mass conservation law and weighting the flow by the PV sectional area. Results: An example of the simulated CFD in SR (middle panels) and AF (right panels) is shown in the figure, coupled with the mitral valve flow-rate computed by the simulation (left panels), which permits to identify the phase of the cardiac cycle. Simulated velocity profile at the mitral valve showed a “physiological” behavior with amplitude compatible with AF condition (mean/peak velocity flow SR: 34 cm;s-1 / 67 cm;s-1; AF: 35 cm;s-1 / 70 cm;s-1). Moreover, in the SR simulation, peak A-wave velocity was 38 cm;s-1 while the A-wave was absent in AF. LA appendage (LAA) blood flow stasis was quantified by counting the number of particles remaining in the LAA throughout the cardiac cycles. Therefore, we placed 500 fluid particles in the LAA at the beginning of the simulation and we found that, after three cardiac cycles, 26% of the particles remained in the LAA in the SR condition, while 45.6% of them remained in the LAA in the AF condition. Conclusion: We developed the first patient-specific CFD model of the LA in AF. Our model includes patient-specific morphology and function and potentially enables optimized patient risk stratification and therapy. In this preliminary testing, the model returned realistic velocity profiles and showed reduced blood washout in the LAA in AF, which might facilitate clot formation. Comparison with a healthy LA for a better understanding of AF hemodynamics as well as further simulations including different LAA morphologies are part of our ongoing research. (Figure presented).