Abstract
Noninvasive prediction of Fractional Flow Reserve (FFR) from imaging data through computational modeling has emerged as a promising alternative to invasive pressure measurements. Simulating coronary physiology, specifically coronary stenosis, poses a significant challenge due to the complex geometries of stenotic lesions and the need for physiologically realistic boundary conditions. Coupled 1D-3D modeling frameworks integrate a global one-dimensional (1D) circulation model with localized three-dimensional (3D) Computational Fluid Dynamics (CFD), enabling dynamic updates of boundary conditions and more accurate hemodynamic simulation. In this study, we couple a global 1D model of the coronary tree and partial systemic circulation with 3D CFD simulations using synthetically generated coronary stenosis geometries. We created three lesion types-symmetric, eccentric, and irregular-at severities of 50%, 70%, and 80%, to evaluate explicit coupling strategies for FFR prediction. We compare a steady-state 3D simulation driven by mean flow from the transient 1D model with transient 3D simulations that exchange data continuously at every step or only at the end of the converged cardiac cycle. Applied to the synthetic stenotic geometries, all approaches predicted similar FFR values, while the steady-state strategy achieved a significant reduction in computational cost, rendering it the most efficient for FFR prediction. Moreover, for irregular lesion geometries, localized 3D modeling revealed discrepancies in pressure loss compared to a simplified lumped model, demonstrating the added value of high-fidelity 3D simulations in complex cases.