A Multi-modal Computational Fluid Dynamics Model of Left Atrial Fibrillation Haemodynamics Validated with 4D flow MRI

Abstract

Atrial fibrillation (AF) is characterized by rapid and irregular contraction of the left atrium (LA). Impacting LA haemodynamics, this increases the risk of thrombi development and stroke. The precise haemodynamic conditions that precede stroke in these patients are not well defined, in part due to a lack of resolution in current 4D flow MRI (magnetic resonance imaging). In this study we combine a high-resolution CT (computed tomography) reconstruction with motion data and inlet flow rates from 4D flow MRI to create a novel multimodal computational fluid dynamics (CFD) model of the LA. Using the 4D flow MRI data we validate the dominant flow structures. Comparing this model with one which assumes rigid walls, a common simplification for such models, shows this assumption has a strong impact on time-averaged wall shear stress (TAWSS) and oscillatory shear index (OSI) in the left atrial appendage (LAA), underestimating these by 61% and 41%, respectively. The dynamic model also yields a more accurate mitral valve outflow when compared with 4D flow MRI measurements. We then applied the model to a cohort of five AF patients prior to catheter ablation, showing in the LAA the low TAWSS and high OSI associated with thrombi growth. We also present a correlation between both LA/LAA volume and pro-thrombotic conditions. Taken together this work demonstrates the importance of LA motion in CFD models that seek to accurately represent LAA haemodynamics, particularly OSI. With further analysis in larger cohorts, LAA OSI may have predictive benefit for stroke in AF patients.