Introduction. A computational fluid-dynamics (CFD) model previously developed with the aim of evaluating cardioembolic risk in patient affected by atrial fibrillation was used for the characterization of the left atrium (LA) hemodynamic and the stratification of the cardioembolic risk in normal sub-jects (NL), patients affected by paroxysmal atrial fibrillation (PAR-AF) and patients affected by persistent atrial fibrillation (PER-AF). Methods. 3D patient-specific anatomical and motion models were derived from ECG-gated coronary artery CTs acquired with retrospective protocol. These models represented the computational domain for CFD simulation in which inflow initial conditions were derived from PW Doppler at the mitral valve and at the pulmonary veins. Velocity field and vortex structures both within the LA and left atrial appendage (LAA) were assessed in 10 NL, 5 PAR-AF and 4 PER-AF. Blood stasis was evaluated by populating the LAA with 500 particles and counting the number of particles still present after five cardiac cycles.
Results. Velocities inside the LA and in the LAA presented different amplitude and distribution in the 3 groups (peak velocity – NL: 50÷60cm/s, PAR-AF: 40÷50cm/s, PER-AF: 15÷25cm/s). The mean velocity resulted lower in the PAR-AF compared with PERS-AF (mean velocity – PAR-AF: 25÷35cm/s, PER-AF: 8÷20cm/s) at the LAA ostium and inside the LAA, in which the wash-out effect was strongly reduced. A higher number of vortex structures was observed in NL compared with AF patients, thus favouring the hypothesis of a more efficient washout of the LA and of the LAA. The fluid particle analysis in the LAA confirmed these results (NL: 5±2, PAR-AF: 18±3, PER-AF: 41±10). Conclusions. The developed approach quantifies differences in LA hemodynamic between AF and NL patients, also allowing a stratification of the disease progression in terms of variations in the blood velocity, organization of blood flow and quantification of blood stasis.