Comparative Evaluation of Carotid Artery Hemodynamics: Patient-Specific CFD Simulations vs. 4D flow MRI
Wall shear stress (WSS) is implicated in carotid atherosclerosis, and 4D flow MRI offers a non-invasive route to its estimation, but the agreement of MRI-derived plaque-surface WSS with computational fluid dynamics (CFD) has not been quantified in a large cohort. We compared peak-systolic velocity and WSS from 4D flow MRI and patient-specific CFD in 240 carotid arteries of 120 patients, with CFD geometries and boundary conditions derived from MRI. Velocity and circumferential WSS were compared at three standardized cross-sections, and plaque-surface WSS in 63 of 149 stenosed arteries with a reliably segmented plaque. Agreement was assessed with Bland-Altman analysis, Spearman correlation with stenosis degree, and exploratory multivariable regression of the plaque-surface discrepancy on hemodynamic and geometric variables. MRI reproduced the broad flow features but underestimated peak velocity, increasingly toward the distal internal carotid artery. WSS disagreement was location dependent: mean circumferential WSS showed little average bias in the common carotid artery, whereas MRI missed the high WSS at the bifurcation apex and underestimated plaque-surface mean and maximum WSS by $40.9 \pm 28.8\%$ and $44.5 \pm 26.8\%$; the difference in Pa increased with stenosis degree ($ρ= 0.47$ and $0.50$). In exploratory regression, the CFD-derived pressure drop across the plaque was the only significant correlate of the mean WSS discrepancy, whereas stenosis degree and geometric variables were associated with the maximum WSS discrepancy. Under the present protocol, MRI-derived plaque-surface WSS showed magnitude- and location-dependent disagreement with CFD, greater in more severely stenosed arteries and, in exploratory analyses, associated with factors beyond stenosis degree.