Modelling blood flow in coronary arteries: Newtonian or shear-thinning non-Newtonian rheology?

ABSTRACT

BACKGROUND: The combination of medical imaging and computational hemodynamics is a promising technology to diagnose/prognose coronary artery disease (CAD). However, the clinical translation of in silico hemodynamic models is still hampered by assumptions/idealizations that must be introduced in model-based strategies and that necessarily imply uncertainty. This study aims to provide a definite answer to the open question of how to properly model blood rheological properties in computational fluid dynamics (CFD) simulations of coronary hemodynamics. METHODS: The geometry of the right coronary artery (RCA) of 144 hemodynamically stable patients with different stenosis degree were reconstructed from angiography. On them, unsteady-state CFD simulations were carried out. On each reconstructed RCA two different simulation strategies were applied to account for blood rheological properties, implementing (i) a Newtonian (N) and (ii) a shear-thinning non-Newtonian (non-N) rheological model. Their impact was evaluated in terms of wall shear stress (WSS magnitude, multidirectionality, topological skeleton) and helical flow (strength, topology) profiles. Additionally, luminal surface areas (SAs) exposed to shear disturbances were identified and the co-localization of paired N and non-N SAs was quantified in terms of similarity index (SI). RESULTS: The comparison between paired N vs. shear-thinning non-N simulations revealed remarkably similar profiles of WSS-based and helicity-based quantities, independent of the adopted blood rheology model and of the degree of stenosis of the vessel. Statistically, for each paired N and non-N hemodynamic quantity emerged negligible bias from Bland-Altman plots, and strong positive linear correlation (r > 0.94 for almost all the WSS-based quantities, r > 0.99 for helicity-based quantities). Moreover, a remarkable co-localization of N vs. non-N luminal SAs exposed to disturbed shear clearly emerged (SI distribution 0.95 [0.93, 0.97]). Helical flow topology resulted to be unaffected by blood rheological properties. CONCLUSIONS: This study, performed on 288 angio-based CFD simulations on 144 RCA models presenting with different degrees of stenosis, suggests that the assumptions on blood rheology have negligible impact both on WSS and helical flow profiles associated with CAD, thus definitively answering to the question "is Newtonian assumption for blood rheology adequate in coronary hemodynamics simulations?".