Importance of Rheology for Carotid Hemodynamics Lee & Steinman On the Relative Importance of Rheology for Image-Based CFD Models of the Carotid Bifurcation Sang-Wook Lee, Ph.D.1, David A. Steinman, Ph.D.1,2 1 Department of Mechanical and Industrial Engineering and 2Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Canada Accepted for publication in J Biomech Eng, 26 Aug 2006 Corresponding author: David A. Steinman, Ph.D. Department of Mechanical & Industrial Engineering University of Toronto 5 King's College Road Toronto, Ontario, Canada M5S 3G8 Phone: 416-978-7781 Fax: 416-978-7753 E-mail:steinman@mie.utoronto.ca Manuscript Preprint Page 1 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Abstract Background: Patient-specific computational fluid dynamics (CFD) models derived from medical images often require simplifying assumptions to render the simulations conceptually or computationally tractable. In this study, we investigated the sensitivity of image-based CFD models of the carotid bifurcation to assumptions regarding the blood rheology. Method of Approach: CFD simulations of three different patient-specific models were carried out assuming: a reference high-shear Newtonian viscosity; two different non-Newtonian (shearthinning) rheology models; and Newtonian viscosities based on characteristic shear rates or, equivalently, assumed hematocrits. Sensitivity of wall shear stress (WSS) and oscillatory shear index (OSI) were contextualized with respect to the reproducibility of the reconstructed geometry, and to assumptions regarding the inlet boundary conditions. Results: Sensitivity of WSS to the various rheological assumptions was roughly 1.0 dyn/cm2 or 8%, nearly seven times less than that due to geometric uncertainty (6.7 dyn/cm2 or 47%), and on the order of that due to inlet boundary condition assumptions. Similar trends were observed regarding OSI sensitivity. Rescaling the Newtonian viscosity based on time-averaged inlet shear rate served to approximate reasonably, if overestimate slightly, non-Newtonian behaviour. Conclusions: For image-based CFD simulations of the normal carotid bifurcation, the assumption of constant viscosity at a nominal hematocrit is reasonable in light of currently available levels of geometric precision, thus serving to obviate the need to acquire patient-specific rheological data. Keywords: computational fluid dynamics, non-Newtonian rheology, carotid bifurcation, wall shear stress, hematocrit. Manuscript Preprint Page 2 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Introduction Local hemodynamics have long been implicated in normal and pathological blood vessel remodeling, in particular the progression of atherosclerosis [1]. The advancement of computer resources, numerical algorithms in computational fluid dynamics (CFD) and image processing techniques has led to detailed investigations of the hemodynamic parameters affecting the progression of disease, based on anatomically realistic geometry [2]. While such image-based CFD provides valuable information for subject-specific hemodynamic parameters, computational modeling of blood flow dynamics often requires assumptions for physical properties and boundary conditions in order to simplify the complex problem. One of the most common assumptions is Newtonian rheology, which implies a linear shear stress-strain rate relationship, i.e., constant viscosity, even though blood is known to have intrinsic rheological characteristics of shear thinning, viscoelasticity in oscillatory flow and thixotropy due to its slow recovery from transient shear change [3,4]. The shear-thinning property of viscosity is considered to be the predominant nonNewtonian characteristic of blood [3,5]. While many studies have been conducted to investigate the effects of shear thinning under physiological flow conditions by considering various constitutive models, the importance of accounting for non-Newtonian rheology remains unclear. Some studies have reported a pronounced effect of shear-thinning [5-8], while others have shown only a modest effect [9-11]. One reason for this is that, as noted by O'Callaghan et al. [12], the effect of non-Newtonian rheology depends explicitly on geometry and flow rate characteristics, and so should be considered individually in each case. Another reason, however, is that it is usually difficult to appreciate whether the effect of non-Newtonian rheology on local hemodynamic pa- Manuscript Preprint Page 3 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman rameters derived from any particular study is significant in the sense suggested by Friedman et al [13], namely, that it would affect conclusions drawn regarding correlations between local hemodynamics and vascular disease. Previously, we assessed the sensitivity of image-based CFD models to inter-scan variability in reconstructed carotid bifurcation geometry and flow rates [14], and to assumptions regarding the inlet velocity boundary conditions [15]. Here we use those findings to assess the relative importance of assumptions regarding blood rheology, as a practical surrogate test of their hemodynamic significance. Methods The “subjects” of this study were three normal or mildly diseased carotid bifurcation geometries previously reconstructed from black blood magnetic resonance imaging (MRI) as detailed by Thomas et al. [14]. In that study, three elderly volunteers were scanned three times at weekly intervals, from which image-based CFD models and flow rate waveforms were reconstructed. For the present study we used, without loss of generality, the lumen geometry and flow rates reconstructed from the first week’s scans. Numerical Simulation For this study, a well-validated, in-house CFD solver for Newtonian fluid flow [16-18], based on quadratic tetrahedral finite elements, was modified to incorporate non-Newtonian rheology using the unconditionally stable splitting approach suggested by Deville et al. [19]. At every time step the variable viscosity was computed based on the computed shear-strain rate and one of the shear-thinning models described below, and then used to update the diffusion matrix. The non- Manuscript Preprint Page 4 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Newtonian solver was validated by calculation of steady flow in a straight pipe using a powerlaw model, and in a 90˚ curved tube using a Carreau-Yasuda model [20], and showed excellent agreement with the analytic solution and experimental data, respectively. To ensure fully developed velocity profiles at the common carotid artery (CCA) inlet and to minimize the influence from outlet boundary condition, straight flow extensions were added at CCA, external (ECA) and internal carotid artery (ICA). No-slip boundary conditions were applied to the rigid walls. Uniform meshes of elements having a nominal edge length of 0.25 mm were used to ensure grid independence and simulations were run for four cycles to damp initial transients, as previously demonstrated by Moyle et al. [15]. Initially, execution times for nonNewtonian simulations were 90% longer than those of the corresponding Newtonian cases, owing to the need to recompute the diffusion matrix at each time step. This was reduced to roughly 20% by pre-computing fixed terms of the diffusion matrix, at the expense of a nearly twofold increase in storage requirements. Rheological models Two different shear-thinning models, having the same high shear-rate viscosity but different behaviors in low shear-rate region (Fig. 1), were employed for this study: (i) Carreau model Along with its variant, the Carreau-Yasuda model, the Carreau model is one of the most popular for representing the shear-thinning property of blood viscosity. Here viscosity is related to shear rate – in practice, the second invariant of the shear rate tensor, γ& – via [ μ (γ& ) = μ ∞ + (μ 0 − μ ∞ )1 + (λγ& )2 Manuscript Preprint Page 5 ] (n −1) / 2 , (1) 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman where μ∞ = 3.5 cP, μ0 = 2.5 cP, λ = 25 sec, and n = 0.25 [21]. (ii) Ballyk model This model, based on a power law with variable parameters, is referred to as a generalized power law model μ (γ& ) = λ (γ& ) γ& n (γ& ) −1 , (2) where ⎡ ⎛ γ& ⎞ ⎛ − b ⎞⎤ ⎟⎟⎥ , λ (γ& ) = μ ∞ + Δμ exp ⎢− ⎜1 + ⎟ exp⎜⎜ ⎝ γ& ⎠⎦ ⎣ ⎝ a⎠ (3a) ⎡ ⎛ γ& ⎞ ⎛ − d ⎞⎤ ⎟⎟⎥ . n (γ& ) = n ∞ + Δn exp ⎢− ⎜1 + ⎟ exp⎜⎜ ⎝ γ& ⎠⎦ ⎣ ⎝ c⎠ (3b) The parameters in Eqs. (3) , obtained by Ballyk et al. through fitting experimental data from various sources, are: μ∞ = 3.5 cP, n∞ = 1.0, Δμ = 25 cP, Δn = 0.45, a = 35.36, b = 2.12, c = 35.36, d = 2.83 [3]. This model yields higher viscosity at low shear rate (< 10-2), and, in the shear-rate region between 50 and 100 s-1 where the Carreau model may underestimate viscosity, exhibits better a fit to experimental data of Walburn and Schneck [22]. Moreover, this model was suggested recently to achieve a better approximation of WSS in low shear-rate regions for coronary artery simulations [23]. Characteristic viscosity Previous studies by Gijsen et al. [20] and Chen and Lu [6] have suggested that rescaling the Newtonian viscosity to a value corresponding to a characteristic shear rate (as opposed to the usual Manuscript Preprint Page 6 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman assumption of a nominal high shear-rate viscosity) predicts non-Newtonian behavior well, even for pulsatile flow. To test this, we carried out Newtonian simulations for the three subjects using, instead of μ = 3.5 cP, viscosities derived from the Carreau model at characteristic shear rates corresponding to each model’s mean and minimum flow rates. These characteristic shear rates were calculated as [24] γ&c = 8V , 3R (4) where V is the inlet mean velocity for the given flow rate, and R is the inlet radius. The resulting shear rates and characteristic viscosities are provided in Table 1. Effect of Hematocrit By virtue of an empirical relationship between viscosity (in cP) and hematocrit (Hct) [25], μ = 1.4175 + 5.878 Hct − 15.98 Hct 2 + 31.964 Hct 3 , (5) the use of a characteristic Newtonian viscosity may be seen to be equivalent to adjusting the hematocrit. As shown in Table 1, the rescaled viscosities at mean and minimum flow rates correspond to hematocrits of about 0.48 and 0.53, respectively. In a recent study, Box et al. [26] measured hematocrits from 89 blood samples, reporting a mean of 0.43, and a range of 0.35 to 0.51. Their average hematocrit corresponds to our nominal Newtonian viscosity of 3.5 cP, and their maximum hematocrit corresponds to our rescaled viscosity based on minimum flow rate. Therefore, to test the effect of hematocrit over a physiologically relevant range, we also carried out simulations at their minimum hematocrit of 0.35, corresponding to a viscosity of 2.9 cP. Manuscript Preprint Page 7 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Data Analysis In order to place all comparisons in the context of the effect of uncertainty in the geometric reconstruction, the surface of each model was divided into “patches”, whose location is fixed with respect to a local bifurcation coordinate system. Patching serves to transform the continuous vascular surface into a finite number of contiguous rectangular regions, over each of which hemodynamic variables of interest (e.g., WSS, OSI) are averaged. This effectively maps these continuous variables onto a template parametric plane, which facilitates quantitative comparisons [27]. Following Thomas et al. [14] and Moyle et al. [15], patches were nominally 1.5-mm long in the direction of the vessel axis, with eight patches distributed around the vessel circumference, resulting in 336, 272 and 216 patches for subjects A, B and C, respectively, in the region of MRI reconstructed data. Note that all analyses excluded both inlet and outlet flow extensions. The effect of rheology on the computed hemodynamics was quantified by calculating the root-mean-square (RMS) of patch-wise difference in time-averaged WSS and OSI relative to CFD simulations carried in the same model under the usual assumption of Newtonian flow with nominal high-shear viscosity of 3.5 cP. Results Fig. 2 demonstrates the negligible differences between time-averaged WSS patterns computed assuming Newtonian vs. non-Newtonian rheology. Mild differences are evident for the rescaled Newtonian models, but these are seen to be appreciably less than those due to differences among the week 1, 2 and 3 reconstructions. These trends are also evident in the patterns of OSI dis- Manuscript Preprint Page 8 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman played in Fig. 3, although differences among the models appear to be somewhat greater than those for WSS. A quantitative comparison of the different models, patched and relative to the reference case shown panels (a) of Fig. 2 and Fig. 3 is summarized in Table 2. The Ballyk model is seen to exhibit slightly more rheological effect than the Carreau model for all subjects, with percentage differences of 8.2 % (Ballyk) and 7.2 % (Carreau) relative to the WSS averaged over all patches of the Newtonian model for the respective subject. Comparing the effect of rheological model to that of geometric uncertainty reveals that the variability of geometry as given inherently by successive scans ranged from four to ten times greater than the effect of the shear-thinning property of viscosity, with variation in geometry inducing changes of the order of 50% of the average WSS. The rescaled Newtonian fluid computations based on mean and minimum flow rates also showed WSS percent differences averaging 9.1 % and 13.7 %, respectively, and in most cases greater than the corresponding non-Newtonian simulations. (When compared instead against Carreau non-Newtonian simulations, these percent differences averaged 11.5% and 15.7%, respectively.) Simulations using a hematocrit of 0.35 showed 7.2% difference overall compared to reference Newtonian flow with a normal hematocrit of 0.43. Similar trends were observed for the differences in OSI among the different models. Discussion In the present study, we carried out numerical computations of pulsatile flow based on three subject-specific carotid bifurcations, using various rheological assumptions to represent the shearthinning properties of blood or typical values of blood hematocrit. It was shown that shear- Manuscript Preprint Page 9 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman thinning viscosity has a relatively minor influence on the hemodynamics of the carotid bifurcation, specifically time-averaged WSS and OSI. This can be attributed to the fact that shear thinning is evident only at low shear rates, whereas relatively high shear rates (typically 250–450 s-1 averaged over the cardiac cycle) are present around carotid bifurcation. The practice of rescaling the constant viscosity was found to be reasonable if based on characteristic shear rates at mean flow rates, but appeared to overestimate the shear-thinning effects when the characteristic shear rate at minimum flow rate was used. Moreover, differences between corresponding rescaled Newtonian and non-Newtonian simulations were marginally higher than differences between corresponding high shear Newtonian and non-Newtonian cases, confirming that rescaling overestimates the non-Newtonian effects. Overall, however, and especially as evidenced in Fig. 2 and Fig. 3, differences among these various rheological models paled in comparison to the differences due to uncertainty in the reconstructed geometry. Our findings are seemingly at odds with those of Gijsen et al. [5], who conducted experimental and numerical simulations of non-Newtonian fluid in a three dimensional carotid bifurcation model with circular cross-sections, and reported considerable differences compared to Newtonian fluid computation. In particular, velocity profiles in the ICA were found to be flattened, with a lower velocity gradient at the divider wall for the non-Newtonian fluid. Curvature-induced secondary flow was found to be less pronounced as well. Chen and Lu [6] also reported a pronounced influence of non-Newtonian pulsatile flow in a non-planar bifurcation model with constant-radius circular cross-section. Shear-thinning flow exhibited flattened axial velocity and higher WSS on the daughter vessels overall. However, WSS distributions were shown to be Manuscript Preprint Page 10 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman closer to Newtonian flow at relatively high shear rates, and to rescaled Newtonian flow in lower shear regions. In the present study as well, it was found that, although the overall rheological effect was small for the time-averaged WSS fields, it was more pronounced on the instantaneous velocity profiles at the diastolic phase when relatively low shear rate was present, as shown in Fig. 4. For the same reason, rescaled Newtonian fluid computations based on mean flow rates showed better prediction of the non-Newtonian effects during diastole. The relatively minor rheological effects on hemodynamics in the carotid bifurcation as shown in the present study relative to those reported by Gijsen et al. and Chen and Lu may be explained by considering some of the idealizations used for those studies. First, the present study is based on anatomically realistic geometries and physiological flow waveforms obtained from MRI of individual subjects. Subject-specific geometries are asymmetric and have irregular surfaces, whereas the circular cross-sections of idealized models generally create axisymmetric flow patterns or organized secondary flow, which exhibit a pronounced shear-thinning effect by flattening the axial velocity profile, and which are inherently more sensitive to perturbations. In addition, the flow waveforms used in the studies of Gijsen et al. and Chen and Lu had a pronounced period of very low diastolic flow, which is not generally seen in vivo. Moreover, the rheological model obtained by fitting experimental data of KSCN-Xanthan gum solution seems to overestimate the blood viscosity within the typical range of shear rate in carotid bifurcation geometry (1– 450 s-1). Our findings regarding the sensitivity of computed hemodynamics to hematocrit were consistent with those of Box et al. [26]. Specifically, inter-subject variations in hematocrit on the Manuscript Preprint Page 11 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman order of ±20 % were found to result in RMS WSS differences ranging from 5-15%, which is comparable to the 10-15 % differences in WSS reported by Box et al. over the same range of hematocrits. As to the question of whether it is important to consider the patient-specific viscosity, we reiterate that RMS differences in WSS and OSI due to geometric uncertainty were nearly six times those due to the choice of hematocrit. To emphasize this point further, the influence of rheology on WSS and OSI was similar to that reported previously by our group regarding the influence of inlet velocity boundary conditions, and indeed the unavoidable physiological “noise” in the measured flow rates themselves [15]. Finally, it is worth noting that the non-Newtonian models used here do not account for the disparate time scales for red blood cell (RBC) aggregation and disaggregation, the physical mechanisms underlying the shear-thinning behavior. RBC aggregation takes much longer than the period of a typical cardiac cycle [28], and so while aggregates may have time to form in persistent local recirculation regions having well-established dividing streamlines, in the carotid sinus recirculatory flow is often largely washed out during systole [29,30]. We expect, therefore, that real blood flow in carotid bifurcation falls somewhere between the extremes of the Newtonian and non-Newtonian simulations carried out here, and so our study likely overestimates the already modest apparent errors incurred under the assumption of Newtonian rheology. Conclusions In summary, our findings suggest that, for image-based CFD simulations of flow in the normal carotid bifurcation, the assumption of Newtonian rheology is reasonable in the context of currently available levels of geometric precision, and assumptions and uncertainties related to the Manuscript Preprint Page 12 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman inlet boundary conditions. This is of practical importance because non-Newtonian fluid simulations can require greater CPU effort and/or storage requirements compared to Newtonian fluid simulations due to the need to recompute the diffusion matrix at every time step, and for patientspecific models it is not always possible or convenient to measure each patient’s viscosity. Therefore, we reiterate the conclusion of our previous study on the effect of inlet boundary condition assumptions [15], namely, that (uncertainty of) the reconstructed geometry has primary influence on physiologically significant hemodynamic indicators. Acknowledgments This work was supported by a grant from the Canadian Institutes of Health Research (MOP62934). DAS was supported by a Career Investigator Award from the Heart & Stroke Foundation of Ontario. All simulations were carried out on the High Performance Advanced Computing Facility (HPACF) at the University of Toronto Institute for Aerospace Studies. References [1] A. M. Malek, S. L. Alper, and S. Izumo, 1999, "Hemodynamic shear stress and its role in atherosclerosis," JAMA, 282, pp. 2035-42. [2] D. A. Steinman, 2002, "Image-based computational fluid dynamics modeling in realistic arterial geometries," Ann Biomed Eng, 30, pp. 483-97. [3] P. D. Ballyk, D. A. Steinman, and C. R. Ethier, 1994, "Simulation of non-Newtonian blood flow in an end-to-side anastomosis," Biorheology, 31, pp. 565-86. Manuscript Preprint Page 13 8/27/2006 Importance of Rheology for Carotid Hemodynamics [4] Lee & Steinman M. K. Sharp, G. B. Thurston, and J. E. Moore, Jr., 1996, "The effect of blood viscoelasticity on pulsatile flow in stationary and axially moving tubes," Biorheology, 33, pp. 185208. [5] F. J. Gijsen, F. N. van de Vosse, and J. D. Janssen, 1999, "The influence of the nonNewtonian properties of blood on the flow in large arteries: steady flow in a carotid bifurcation model," J Biomech, 32, pp. 601-8. [6] J. Chen and X. Y. Lu, 2006, "Numerical investigation of the non-Newtonian pulsatile blood flow in a bifurcation model with a non-planar branch," J Biomech, 39, pp. 818-32. [7] W. L. Siauw, E. Y. Ng, and J. Mazumdar, 2000, "Unsteady stenosis flow prediction: a comparative study of non-Newtonian models with operator splitting scheme," Med Eng Phys, 22, pp. 265-77. [8] C. Tu and M. Deville, 1996, "Pulsatile flow of non-Newtonian fluids through arterial stenoses," J Biomech, 29, pp. 899-908. [9] K. Perktold, M. Resch, and H. Florian, 1991, "Pulsatile non-Newtonian flow characteristics in a three-dimensional human carotid bifurcation model," J Biomech Eng, 113, pp. 464-75. [10] Y. I. Cho and K. R. Kensey, 1991, "Effects of the non-Newtonian viscosity of blood on flows in a diseased arterial vessel. Part 1: Steady flows," Biorheology, 28, pp. 241-62. Manuscript Preprint Page 14 8/27/2006 Importance of Rheology for Carotid Hemodynamics [11] Lee & Steinman A. Valencia, A. Zarate, M. Galvez, and L. Badilla, 2006, "Non-Newonian blood flow dynamics in a right internal carotid artery with a saccular aneurysm," Int J Num Meth Fluids, 50, pp. 751-764. [12] S. O'Callaghan, M. Walsh, and T. McGloughlin, 2006, "Numerical modelling of Newtonian and non-Newtonian representation of blood in a distal end-to-side vascular bypass graft anastomosis," Med Eng Phys, 28, pp. 70-4. [13] M. H. Friedman, C. B. Bargeron, D. D. Duncan, G. M. Hutchins, and F. F. Mark, 1992, "Effects of arterial compliance and non-Newtonian rheology on correlations between intimal thickness and wall shear," J Biomech Eng, 114, pp. 317-20. [14] J. B. Thomas, J. S. Milner, B. K. Rutt, and D. A. Steinman, 2003, "Reproducibility of image-based computational fluid dynamics models of the human carotid bifurcation," Ann Biomed Eng, 31, pp. 132-41. [15] K. R. Moyle, L. Antiga, and D. A. Steinman, 2006, "Inlet conditions for image-based CFD models of the carotid bifurcation: Is it reasonable to assume fully-developed flow?" J Biomech Eng, 128, pp. 371-9. [16] C. R. Ethier, D. A. Steinman, and M. Ojha, "Comparisons between computational hemodynamics, photochromic dye flow visualization and magnetic resonance velocimetry," in The Haemodynamics of Arterial Organs-Comparisons of Computational Predictions with Manuscript Preprint Page 15 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman in vivo and in vitro Data, X. Y. Xu and M. W. Collins, Eds.: WIT Press, 1999, pp. 131183. [17] P. D. Minev and C. R. Ethier, 1999, "A characteristic/finite element algorithm for the 3-D Navier-Stokes equations using unstructured grids," Comp Meth App Mech Eng, 178, pp. 39-50. [18] C. R. Ethier, S. Prakash, D. A. Steinman, R. L. Leask, G. G. Couch, and M. Ojha, 2000, "Steady flow separation patterns in a 45 degree junction," J Fluid Mech, 411, pp. 1-38. [19] M. O. Deville, P. F. Fischer, and E. H. Mund, High order methods for incompressible fluid flow: Cambridge University Press, 2002. [20] F. J. Gijsen, E. Allanic, F. N. van de Vosse, and J. D. Janssen, 1999, "The influence of the non-Newtonian properties of blood on the flow in large arteries: unsteady flow in a 90 degrees curved tube," J Biomech, 32, pp. 705-13. [21] T. Seo, L. G. Schachter, and A. I. Barakat, 2005, "Computational study of fluid mechanical disturbance induced by endovascular stents," Ann Biomed Eng, 33, pp. 444-56. [22] F. J. Walburn and D. J. Schneck, 1976, "A constitutive equation for whole human blood," Biorheology, 13, pp. 201-10. Manuscript Preprint Page 16 8/27/2006 Importance of Rheology for Carotid Hemodynamics [23] Lee & Steinman B. M. Johnston, P. R. Johnston, S. Corney, and D. Kilpatrick, 2004, "Non-Newtonian blood flow in human right coronary arteries: steady state simulations," J Biomech, 37, pp. 709-20. [24] G. B. Thurston, 1973, "Frequency and shear rate dependence of viscoelasticity of human blood," Biorheology, 10, pp. 375-81. [25] A. C. Guyton, Medical Physiology, 6th ed. Philadelphia: W.B.Sounders Company, 1981. [26] F. M. Box, R. J. van der Geest, M. C. Rutten, and J. H. Reiber, 2005, "The influence of flow, vessel diameter, and non-Newtonian blood viscosity on the wall shear stress in a carotid bifurcation model for unsteady flow," Invest Radiol, 40, pp. 277-94. [27] L. Antiga and D. A. Steinman, 2004, "Robust and objective decomposition and mapping of bifurcating vessels," IEEE Trans Med Imaging, 23, pp. 704-13. [28] G. R. Cokelet, 1980, "Rheology and hemodynamics," Annu Rev Physiol, 42, pp. 311-24. [29] R. S. Reneman, T. van Merode, P. Hick, and A. P. Hoeks, 1985, "Flow velocity patterns in and distensibility of the carotid artery bulb in subjects of various ages," Circulation, 71, pp. 500-9. [30] D. N. Ku and D. P. Giddens, 1987, "Laser Doppler anemometer measurements of pulsatile flow in a model carotid bifurcation," J Biomech, 20, pp. 407-21. Manuscript Preprint Page 17 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Table 1. Carreau model viscosities and equivalent hematocrits based on characteristic inlet wall shear rates at mean ( γ&c ,mean ) and minimum ( γ&c ,min ) flow rates. γ&c ,mean (s-1) μ c ,mean (cP) Hct γ&c ,min (s-1) μ c ,min (cP) Hct Subject A 117 4.1 0.48 47 4.7 0.52 Subject B 124 4.1 0.48 39 4.9 0.54 Subject C 151 4.0 0.47 56 4.6 0.52 Manuscript Preprint Page 18 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Table 2. RMS differences in time-averaged wall shear stress (WSS) and oscillatory shear index (OSI), relative to the respective Week 1 model with a constant viscosity of 3.5 cP. Rheological influence Variable viscosity WSS (dyn/cm2) Constant viscosity Geometric influence (From ref. [15]) Carreau Ballyk μc,mean μc,min Hct = 0.35 Week 2 Week 3 Subject A 0.87 0.97 0.70 1.09 1.10 3.33 3.84 Subject B 0.90 0.93 1.30 2.20 0.94 11.5 6.79 Subject C 1.20 1.52 2.31 3.09 0.66 7.42 11.2 Average* 0.97 1.10 1.32 1.98 0.93 7.10 6.74 Subject A 8.65 9.64 6.96 10.8 10.9 32.6 37.6 Subject B 5.03 5.19 7.26 12.3 5.25 64.7 38.2 Subject C 7.70 9.76 14.8 19.8 4.24 47.2 71.4 Average* 7.21 8.20 9.11 13.7 7.21 47.0 46.7 Subject A 8.35 13.3 11.1 15.9 9.51 35.0 36.7 Subject B 3.69 5.77 5.53 9.01 4.45 38.2 28.3 Subject C 10.8 13.5 17.3 18.7 11.6 79.8 76.1 Average* 7.45 10.7 10.9 14.4 8.39 47.8 44.3 WSS (%) OSI (×10-3) *Average weighted by the number of patches in each model. Manuscript Preprint Page 19 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Fig. 1. Blood viscosity as a function of shear rate for the different rheological models. Manuscript Preprint Page 20 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Fig. 2. Distribution of time-averaged WSS, based on the Week 1 geometry and assuming: (a) constant viscosity of 3.5 cP; (b) Carreau and (c) Ballyk viscosity models; rescaled Newtonian viscosity derived from each subject’s characteristic inlet shear rate at (d) mean and (e) minimum flow; (f) Newtonian viscosity assuming Hct = 0.35. Also shown are WSS patterns based on (g) Week 2 and (h) Week 3 models assuming a Newtonian viscosity of 3.5 cP. Manuscript Preprint Page 21 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Fig. 3. Distribution of oscillatory shear index (OSI). Refer to caption of Fig. 2 for details. Manuscript Preprint Page 22 8/27/2006 Importance of Rheology for Carotid Hemodynamics Lee & Steinman Fig. 4. Comparisons of axial velocity profiles at the nominal midplane at peak systole and end diastole for the Newtonian model with μ = 3.5 cP (thick solid line), the Carreau model (thin solid line) and the rescaled Newtonian model based on characteristic viscosity at mean flow rate, μc,mean (thin dashed line). Note that velocities are normalized by bulk velocity at each location. Manuscript Preprint Page 23 8/27/2006