Comparison of blood viscosity models in different degrees of carotid artery stenosis

Carotid atherosclerotic vascular disease significantly contributes to strokes, presenting a heightened risk of early recurrent ischemia. Computational fluid dynamics (CFD) has shown potential in predicting subsequent stroke recurrence in patients with carotid stenosis.

This study aims to investigate the differences in computational time and accuracy of four key hemodynamic indices-wall shear stress (WSS), time-averaged wall shear stress (TAWSS), Oscillatory Shear Index (OSI), and relative residence time (RRT)-across different viscosity models, thereby providing optimal model selection for clinical cases and offering guidance for clinicians' decision-making.

A three-dimensional vessel model was established using computed tomography angiography (CTA), followed by CFD simulations to calculate WSS, TAWSS, OSI, and RRT. The accuracy of the simulations was validated by comparing the results with those from Razavi et al. (10.1016/j.jbiomech.2011.04.023). Numerical errors in different parameters under varying stenosis levels and viscosity models were analyzed.

In the transient state, when degree of stenosis is 38%, 72%-84%, the performance difference between the two is less than 6%. For TAWSS, the difference is 0% when degree of stenosis is 12%, 18%, 26%, 54%, and 76%. For OSI, the difference is 0% when stenosis is 16%, 18%, 26%. For RRT, the difference between the two is 0% when degree of stenosis is 18% and 84%. WSS exhibited an increasing trend with higher degrees of stenosis. TAWSS demonstrated significant variation in moderate to severe stenosis, while OSI increased markedly in cases of moderate to severe stenosis. High RRT values in severely stenotic regions indicated a propensity for atherosclerotic lesion development.

This study systematically quantified the discrepancies between Newtonian and non-Newtonian blood viscosity models in predicting hemodynamic parameters across different degrees of carotid artery stenosis. Statistical analyses revealed significant differences between the two models in WSS, TAWSS, OSI, and RRT (p < 0.001 for all parameters). Newtonian models, while computationally efficient, overestimated TAWSS, OSI, and RRT while underestimating WSS, particularly in moderate to severe stenosis. In contrast, non-Newtonian models provided more physiologically accurate predictions, especially in regions with high shear stress variations. The results emphasize the importance of selecting appropriate viscosity models for CFD-based patient-specific risk assessment, particularly in stroke prediction, plaque evaluation, and surgical planning. Non-Newtonian models should be prioritized in high-risk patients where flow disturbances are more pronounced, whereas Newtonian models remain suitable for early screening and rapid assessments.