Heart failure is a common cardiovascular disease globally.1 Because of the permanent shortage of donor hearts for transplantation, ventricular assist devices (VADs) have become an alternative treatment option for these patients.2 In light of pulsatile blood flow’s benefit for myocardial recovery, perfusion of coronary arteries and end organs, pulsatile VADs are still widely used as paracorporeal mechanical circulatory support devices in clinical applications, especially in pediatric heart failure patients.3–53–53–5
However, adverse events associated with VAD blood damage usually occur in clinical applications.6–86–86–8 The blood damage of VADs includes not only hemolysis but also thrombosis. Increased levels of free red blood cell constituents together with an exhaust of their scavengers result in a variety of serious clinical sequelae, such as increased systemic and pulmonary vascular resistance, altered coagulation profile, platelet dysfunction, renal tubular damage, and multiple organ failure.8–108–108–10 Thrombosis can lead to cerebral microembolization and neurological events, heart attacks, and impairment of kidney and liver function.11
In clinic applications, the VADs usually work at off-design conditions, which deteriorates the adverse events associated with VAD blood damage.6,8 Blood damage in VAD is mainly caused by nonphysiologic blood flow patterns, such as high shear stress, long retention time, recirculation, and flow stagnation.11–1311–1311–13 Because operating conditions of a pulsatile VAD may have an influence on the blood flow patterns,14 investigating different operating conditions of VAD may potentially reduce blood damage and also provide references for how to change anticoagulation therapy. In this aspect, Nanna et al.14 and Oley et al.15 investigated the influence of pulsatile rate and systolic duration on pulsatile VAD’s wall shear rate and thrombus deposition. Cooper et al.16 studied the effect of end-diastolic delay time on inner flow pattern of a pediatric pulsatile VAD. These studies indicated that higher pulsatile rate and shorter diastolic time can decrease the potential of thrombus deposition. Shahraki and Oscuii17 compared three different driver patterns (linear, sinusoidal, and Guyton’s pulse) to minimize shear stress–induced blood damage and found that the sinusoidal pattern was superior. Additionally, Navitsky et al.18,19 investigated the thrombus susceptibility potential (TSP) for comparisons of designs and flow waveforms for platelet adhesion, and the results indicated that the design and flow waveform that reduced the region of low shear rate can decrease the potential of thrombosis. These investigations suggest that the blood damages of pulsatile VADs driven by pneumatic driver, linear motor, or maglev pusher plate are all influenced by operating conditions, such as driving patterns, pulsatile rate, and systolic duration.
In previous investigations, however, some factors such as stroke volume (SV) and assist mode,20,21 were not included, and only one aspect of blood damage was focused. Moreover, most of the investigational results were just qualitative. To obtain a comprehensive understanding of the blood damage effect, in this article, a pulsatile VAD was chosen to investigate the influence of different operating conditions on its blood damage. The blood damage indices included hemolysis, platelet activation, and platelet deposition and their values were calculated by fluid–structure interaction (FSI) simulations. An in vitro hemolysis experiment was performed to partially validate the computational model. Three motion profiles of pusher plate (sine, cosine, and polynomial), three SVs (ejection fractions) (56 ml [70%], 42 ml [52.5%], and 28 ml [35%]), three pulsatile rates (75, 100, and 150 bpm), and two assist modes (copulsation and counterpulsation) were implemented respectively in the FSI simulations to calculate the corresponding blood damage indices, and their effects on blood damage were analyzed and discussed.
Materials and Methods
The Pulsatile VAD
The pulsatile VAD in this study is a positive displacement blood pump that its sac bottom is bonded with a pusher plate, and the plate’s reciprocating motion causes blood sac’s systole and diastole. Two bileaflet mechanical valves are installed at inlet/outlet ports to ensure the unidirectional blood flow. The VAD’s inlet diameter is 13 mm and outlet diameter is 17 mm. The inlet tube is tangent to the chamber, whereas the outlet tube perpendicularly locates at the center of the chamber’s top face. The sac is made of silicone rubber, the upper part (cover part) of the chamber is made of titanium alloy, and both of them constitute the blood chamber. The computational domain of the blood chamber in FSI simulation is shown in Figure 1. To minimize the end effect of the openings, extended regions were incorporated at inlet/outlet port of the computational model.
Blood Damage Index
In this article, three mathematical models of blood damage associated with hemocompatibility were chosen as the evaluation indices.
Hemolysis model. According to the work of Taskin et al.,22 a power law model for erythrocyte lysis (Equation 1) was used to evaluate the hemolysis because the predicted hemolysis value with this model was closer to the experimental hemolysis value:
in which τ is the scalar shear stress (shown in Equation 2), and
is the erythrocyte’s exposure time under τ.
is shear stress tensor.22,23 Hemolysis index (IH) is calculated in one full pulsatile cycle.
Platelet activation model. A platelet activation index ( Equation 3) used by Girdhar et al.23 and Taskin et al.24 was applied in this work to assess the level of platelet activation:
in which τ is the scalar shear stress (shown in Equation 2) and
is the platelet’s exposure time under τ. Dpl is calculated in one full pulsatile cycle.
Platelet deposition model. Based on animal experiments, Medvitz12,25 proposed an index entitled TSP describing the thrombus susceptibility potential caused by platelet deposition on the pulsatile VAD’s blood contact surface. The TSP equation is given as follows:
is the scalar wall shear rate.
are two critical values of wall shear rate. And texposure is an exposure time needed per second of pump operation to prevent deposition at
. N is the number of time steps per cycle and Δt is the time step. The value of each constant in Equation 4 is chosen according to previous experiments,12,13,26 as shown in Table 1.
Based on the proposed TSP in Equation 4, an index (Total TSP) denoting the percentage of the thrombosis area in the whole blood contact surface is defined by Equation 5:
denotes the integral of TSP on the chamber surface. Lower Total TSP means less potential of thrombosis in the blood chamber. The Total TSP calculation was conducted on the titanium alloy cover part and the silicone rubber sac and pusher plate.
Fluid–structure interaction simulation of the pulsatile VAD was implemented using the two-way FSI module—System Coupling in ANSYS workbench environment (ANSYS Inc., Canonsburg, PA), in which the fluid solver was ANSYS FLUENT (ANSYS Inc.), the structural solver was ANSYS Mechanical (ANSYS Inc.), and the data transferring between them were controlled by System Coupling component.
To ensure a reasonable CPU time and appropriate computational accuracy, a grid independency check (Table 2) for the computational domain was performed. The relative differences of the blood damage indices between the model of 2.2 million elements and the model of 3.8 million elements were less than 0.3%, which indicated that 2.2 million tetrahedral elements for the fluid domain and 0.12 million tetrahedral elements (including 10 nodes in each element) for the structure domain were appropriate. In fluid domain, five layers of prismatic elements were used in the fluid–solid interface layer and the vicinity of walls. Both of the meshes were generated in ANSYS Meshing (ANSYS Inc.).
Blood was modeled as Newtonian fluid with a density of 1,040 kg/m3 and a viscosity of 3.5 cP according to the shear range under consideration.23 The method of Lagrangian particle tracking was used to calculate the values of IH and Dpl. Particles were released at inlet face during the whole diastole phase. Sixty thousand, 120,000 and 240,000 particles were tested, and the relative differences of the blood damage indices between 120,000 particles and 240,000 particles were less than 0.5%, which indicated that 120,000 particles were appropriate to calculate the values of IH and Dpl. The sac is made of silicone rubber, whose density is 1,100 kg/m3, with a Young’s modulus of 20 MPa and a Poisson’s ratio of 0.49.27 The stiffness behavior of the sac was defined as being flexible.
The bileaflet valves were modeled using the method similar to Medvitz.28 The valves were fixed in the fully opened phase. To model the valve closure phase, the fluid viscosity in the region local to the valves was increased by 1e+4, resulting in a velocity of approximating zero through the “closed” valve. Using the method similar to Long et al.,3 the afterload pressure at outlet was set to 80–120 mm Hg to mimic the output pressure fluctuation of pulsatile VAD. The preload at inlet was set to 10 mm Hg. A fixed support boundary condition was imposed at the sac’s margin. A velocity profile determined by the pusher plate motion was imposed on the bottom flat of the sac because it was bonded to the pusher plate. A fluid–solid interface boundary condition was defined at the upper surface of the sac in the structural domain and the counterpart surface at the bottom of the fluid domain.
Based on the inlet diameter and flow velocity, the Reynolds number can be calculated. During the pulsatile cycle, the peak Reynolds number was 4,985. Therefore, the flow in the blood chamber was turbulent. The realized κ-ε turbulence model was adopted in simulation.11
With the use of Lagrangian formulation in the structural model and arbitrary Lagrangian–Eulerian formulation in the fluid counterpart, the FSI simulation was solved using System Coupling module in ANSYS workbench environment. In light of transient simulation, the time steps of 5 ms, 3 ms, 2 ms, and 1 ms were tested respectively in the simulation. Because the relative differences of the blood damage indices between the time step of 2 ms and 1 ms were less than 0.5%, 1 ms was chosen as the simulation time step, that is, both of the FLUENT and mechanical solver time steps were set to 1 ms. The converged criteria of the residual values were set to 1e-4.11 The pulsatile cycle time was 0.8 s. To avoid the influence of initial conditions, the simulation was run for three cycles because the blood damage indices’ relative differences between the third cycle and the fourth one were less than 0.2%. All data were extracted from the third cycle.
In Vitro Hemolysis Experiment
In this work, an in vitro hemolysis experiment was performed to validate the above computational model. The reason is that in vitro hemolysis experiment for VAD has standard test procedure,22 in which the same measurement index can be conveniently compared with the hemolysis index calculated by simulation, whereas the platelet activation and deposition experiments do not have standard test procedures.
In the hemolysis experiment, blood was freshly collected from the heart of anesthetized adult New Zealand White Rabbits into a 200 ml disposable blood bag (Nigale Biomedical Co., Ltd, Sichuan, China), which contained CPDA-1 (Citrate-Phosphate- Dextrose-Adenosine) preservative. The blood was then introduced into the recirculating flow loop as shown in Figure 2. Before using, the loop chamber and tube had been cleaned by ultrasonic cleaner and the blood-contacting surfaces had been rinsed with normal saline. Air in the loop was removed. The VAD was operated at 75 bpm with an afterload pressure of 120–80 mm Hg and a mean preload pressure of 10 mm Hg, generating an output flow of about 4.2 L/min. The pressure was measured by pressure sensors (PRESS-S-000, PendoTECH, Princeton, NJ), and the flow rate was measured by an ultrasonic flowmeter (MA-16PAU, Transonic, Inc., Ithaca, NY). Blood samples were taken just before VAD operating and every 30 minutes after that. Two milliliters of blood was sampled each time for analysis. The experiment was carried out over a 3 hour period at the room temperature of 25°C. To elevate the measurement accuracy, the experiment was carried out three times under the same operating conditions. The hemolysis induced by VAD was determined by measuring the plasma free hemoglobin with an Ultrospect 2100pro UV/Visible Spectrophotometer (GE Healthcare, Freiburg, Germany) and subsequently calculating the IH22:
where Q is the flow rate (L/min), T is the time that the blood has been circulating (min), and V is the volume of blood in the loop (L).
is the increase of plasma free hemoglobin concentration (g/100 L) over the circulating time T, and
is the hemoglobin concentration of whole blood (g/100 L).
was determined by the slope of
versus T obtained from a linear best-fit of experimental data.
In this work, four influencing factors were studied: three velocity profiles of pusher plate motion (sine, cosine, and polynomial), three SVs (ejection fractions) (56 ml [70%], 42 ml [52.5%], and 28 ml [35%]), three pulsatile rates (75, 100, and 150), and two assist modes (copulsation and counterpulsation) were implemented respectively in the VAD FSI simulations to calculate the corresponding blood damage indices. When one factor such as the motion profile of pusher plate was investigated, the remaining factors including SV, pulsatile rate, and assist mode kept unchanged, thus the effect of this factor on blood damage would be obtained. All the influencing factors and the corresponding unchanged factors are listed in Table 3.
The three velocity profiles of pusher plate motion in Table 3 are plotted in Figure 3.
The variation of SV was implemented by changing the displacement range of the pusher plate.
For the two assist modes in Table 3, copulsation denotes that the preload pressure of the VAD is 10 mm Hg, the afterload pressure is set to 80–120 mm Hg to mimic the output pressure fluctuation of pulsatile VAD,3 and the systolic/diastolic duration is 0.3 s/0.5 s; the counterpulsation denotes that the preload pressure is set to 5–55 mm Hg corresponding to diseased left ventricle pressure20 (varying according to sinusoidal curve with respect to time in the simulation), the afterload pressure is 70–110 mm Hg (assuming that the aortic pressure of the patient reaches a mean value of 90 mm Hg29 after VAD assist), and the systolic/diastolic duration is 0.5 s/0.3 s.
Validation of the Computational Model
The computational model in this work was validated by comparing the predicted and experimental IH value. The predicted IH value was calculated by FSI simulation and the hemolysis model in Equation 1. The boundary conditions in simulation were set according to the hemolysis experiment, with the sine profile of the pusher plate. The predicted IH is 1.162e-4 (%).
The mean value and standard deviation of the experimental IH are shown in Table 4. The relative error between the experimental IH value (1.674e-4 [%]) and the value predicted by numerical simulation (1.162e-4 [%]) is 31%, which may result from the extra hemolysis induced by recirculating tube, experimental error, hemolysis predicted model, or numerical simulation model, but it is still in reasonable range according to the work of Taskin et al.22 The result indicates that the above computational model is partially validated by the in vitro hemolysis experiment, which provides the feasibility of using the computational model to investigate the influence of different operating conditions on blood damage.
The Influence of Different Operating Conditions on the Blood Damage
Figure 4 illustrates the effect of three pusher plate’s motion profiles on VAD blood damage. It is shown that the sine motion’s hemolysis IH and platelet activation Dpl are the lowest, the polynomial’s take the second place, and the cosine’s are the highest, whereas the trend of platelet deposition Total TSP is opposite, decreasing from sine’s 17.82% to cosine’s 4.327%. From Figure 3, it is seen that the velocity range of plate motion gradually increases as the profile varies from sine, polynomial to cosine, resulting in elevated velocity variation rate which consequently yields higher shear stress and wall shear rate, namely, making IH and Dpl higher and Total TSP lower. The results also suggest that the three blood damage indices are conflicting, that is, they cannot be reduced simultaneously by changing one operating condition. The detailed data of the blood damage indices can be found in Table A in Supplemental Digital Content 1 (http://links.lww.com/ASAIO/A67), and the velocity vector plots of the VAD corresponding to three motion profiles can also be seen in Figure A in Supplemental Digital Content 1 (http://links.lww.com/ASAIO/A67).
Figure 5 illustrates the effect of different SVs on VAD blood damage. Different SVs correspond to different ejection fractions. As shown in Figure 5, for the same blood chamber, smaller SV leads to lower IH and Dpl and higher Total TSP. Smaller SV made the velocity in the chamber smaller and vary more gently and subsequently resulted in lower shear stress and worse wall washout, so less hemolysis and platelet activation and more platelet deposition occurred. Moreover, the SV has a great impact on the blood damage indices, for example, Total TSP increases from 56 ml’s 17.82% to 28 ml’s 86.19%.
Figure 6 illustrates the effect of different pulsatile rates on VAD blood damage. As the pulsatile rate varies from 75 to 150 bpm, IH and Dpl increase rapidly and Total TSP decreases sharply. As pulsatile rate increased, the velocity of the blood flow in the chamber varied more drastically during one pulsatile cycle, and the elevated gradient of velocity resulted in pretty high shear stress and adequate wall washout. Therefore, the shear-induced hemolysis and platelet activation increased rapidly, but the platelet deposition reduced obviously.
Figure 7 illustrates the effect of copulsation and counterpulsation on VAD blood damage. As seen from Figure 7, the values of IH and Dpl corresponding to counterpulsation are a little higher than that corresponding to copulsation, whereas the value of Total TSP decreases from 17.82% to 0.2037%. Therefore, the effect of counterpulsation on blood damage is slightly elevating the hemolysis and platelet activation (still remaining at a low level compared with Shahraki and Oscuii17) and significantly decreasing the platelet deposition.
In this work, blood damage differences of a pulsatile VAD under varied operating conditions are revealed by using FSI simulation. Variances are observed in hemolysis, platelet activation, and platelet deposition. The simulation results indicate that the blood damage indices are significantly affected by the operating conditions. And the blood damage indices indicate that cosine motion profile, higher SV (ejection fraction), higher pulsatile rate, and counterpulsation can decrease the potential of platelet deposition whereas increase the potential of hemolysis and platelet activation.
The in vitro hemolysis experiment shows that the studied pulsatile VAD’s IH is 1.674e-4 (%), which is comparable with CentriMag VAD’s IH at 3,000 rpm and 5 L/min and is lower than its IH at 4,000 rpm and 5 L/min.22 Fraser et al.30 studied the hemolysis indices of three clinical VADs and two investigational VADs at 3 L/min and 100 mm Hg, and the IH values of CentVAD1 and CentVAD2 were approximately 2.0e-4 (%) while the IH values of AxVAD1–AxVAD3 were all higher than 5.0e-4 (%). These comparisons indicate that the hemolysis index measured in our work is acceptable, and it also partially validates the computational model.
Shahraki and Oscuii17 compared three different driver patterns of the pusher plate (linear, sinusoidal, and Guyton’s pulse) to reduce the shear stress–induced blood damage. The results indicated that the three types of motion may induce less than 0.06% red blood cell damage, and the sinusoidal pattern caused less hemolysis compared with the other patterns. Additionally, the sinusoidal pattern reduced the probability of thrombosis in comparison to the linear pattern. In our work, among the three motion profiles of pusher plate (sine, polynomial, and cosine), the sine one caused the least hemolysis and platelet activation, which agreed with the results of Shahraki and Oscuii,17 but the corresponding platelet deposition potential in our work was the highest among the three motion patterns. In other words, from the perspective of surface thrombus deposition, the cosine pattern may be more beneficial to reduce blood damage because it yielded a blood flow of higher pulsatility due to higher peak velocity of pusher plate (Figure 3) to minimize the areas of stagnation and low wall shear rate.17,31
Jarvis et al.32 studied the dependence of hemolysis on heart rate and systolic duration using a 100 cc artificial ventricle. They found that the hemolysis was a function of the operating conditions, and higher heart rates and longer systolic durations induced more hemolysis. Oley et al.15 and Nanna et al.14 used particle image velocimetry (PIV) to investigate the influence of pulsatile rate (60–150 bpm) and systolic duration (35–50%) on a 50 cc pulsatile VAD’s wall shear rate and thrombus deposition. They found that diastolic flow was dominated by the inflow jet and showed more penetration and better wall washing at higher beat rates and shorter diastolic time. From the results of our work, it is seen that larger SV (Figure 5), higher pulsatile rate (Figure 6), and shorter diastolic time (Figure 7) will elevate hemolysis and platelet activation and decrease platelet deposition, which are consistent with the above experimental results.
For the VAD assist modes, many studies compared the different hemodynamic effects of copulsation and counterpulsation on cardiovascular system by means of lumped parameter models,29 computational fluid dynamics (CFD) simulations,33 or in vivo experiments.34 These investigations used the indices of energy equivalent pressure, surplus hemodynamic pressure, assistance ratio, and stroke work to evaluate the different effects of the assist modes. Lim et al.29 reported that counterpulsation mode of bypass VAD gave the most physiologic coronary blood perfusion and resulted in a lower peak pressure of left ventricle than other modes, aiding cardiac recovery by reducing the ventricular afterload. However, few investigations associated with the influence of assist modes on VAD blood damage were conducted. In our work, the simulation results suggest that different assist modes affect the blood damage in VAD: different assist modes lead to different preload/afterload pressures and systolic/diastolic durations, and the variation of the latter directly results in the change of blood flow velocity in chamber. Shorter diastolic duration yields higher inlet flow velocity, which is beneficial to reduce the stagnation area in chamber and elevate wall shear rate, that is, reducing the potential of platelet deposition on the chamber wall, but the side effect is making the shear stress higher and yielding more hemolysis and platelet activation during diastole phase. Meanwhile, longer systolic duration is in favor of decreasing the shear stress near the outlet. Finally, the effect of counterpulsation on blood damage is slightly elevating the hemolysis and platelet activation and significantly decreasing the platelet deposition. By combining the effect of significantly decreasing platelet deposition and the effect of enhancing coronary blood perfusion and reducing the ventricular afterload, counterpulsation mode may be superior to copulsation mode in pulsatile VAD’s clinical applications.
For a pulsatile VAD in clinical applications, due to different assist flow rates of the patients, the VAD’s pulsatile rates or SVs are various and the corresponding blood damages are also different, so the effect of pulsatile rate and SV on blood damage may provide references for how to change anticoagulation therapy; under the same assist flow rate, choosing better motion profile of the pusher plate or assist mode can reduce the blood damage of the VAD.
The results shown in Figures 4–7Figures 4–7Figures 4–7Figures 4–7 also reveal that the three blood damage indices are conflicting, that is, the change of Total TSP is always opposite to that of IH and Dpl with varying operating conditions such as motion profiles, SV, pulsatile rate, or assist mode. This also indicates that a balance among the three blood damage indices should be achieved when determining the operating conditions.
Limitations in this study: the FSI simulations in this work excluded valve motion modeling, which may reduce the accuracy of the results; pneumatic VADs have no pusher plate, so the conclusions drawn in this article may be not fully applicable to this type of pulsatile VAD. In the simulation of assist modes, the copulsation and counterpulsation modes were simulated by imposing corresponding boundary conditions on VAD inlet/outlet, and in the future work, the CFD simulation containing the models of left ventricle, aorta, and VAD will be implemented to help analyzing the effect of assist modes on their inner flow fields.
Different operating conditions strongly influence the blood damage of pulsatile VADs. In current study, a pulsatile VAD was used to investigate the influence of different operating conditions on its blood damage, including hemolysis, platelet activation, and platelet deposition. Three motion profiles of pusher plate, three SVs, three pulsatile rates, and two assist modes were implemented respectively in the VAD FSI simulations to calculate the corresponding blood damage. The blood damage indices indicate that cosine motion profile (compared with sine and polynomial), higher ejection fraction, higher pulsatile rate, and counterpulsation (compared with copulsation) can decrease the potential of platelet deposition whereas increase the potential of hemolysis and platelet activation, and vice versa. The results suggest that different operating conditions have obvious effects on pulsatile VAD’s blood damage, and three blood damage indices cannot be reduced simultaneously. The conclusions may be favorable to choose suitable operating condition for pulsatile VADs to diminish blood damage and may also provide references for how to change anticoagulation therapy according to different operating conditions in clinical applications.
1. Lloyd-Jones D, Adams RJ, Brown TM, et al. Heart disease and stroke statistics—2010 update: A report from the American Heart Association. Circulation. 2010;121:e46––e215
2. Miller LW. Left ventricular assist devices are underutilized. Circulation. 2011;123:1552–1558 discussion 1558.
3. Long CC, Marsden AL, Bazilevs Y. Shape optimization of pulsatile ventricular assist devices using FSI to minimize thrombotic risk. Comput Mech. 2014;54:921–932
4. Karimova A, Pockett CR, Lasuen N, et al. Right ventricular dysfunction in children supported with pulsatile ventricular assist devices. J Thorac Cardiovasc Surg. 2014;147:1691.e1–1697.e1
5. Imamura T, Kinugawa K, Hatano M, et al. Preoperative beta-blocker treatment is a key for deciding left ventricular assist device implantation strategy as a bridge to recovery. J Artif Organs. 2014;17:23–32
6. Starling RC, Moazami N, Silvestry SC, et al. Unexpected abrupt increase in left ventricular assist device thrombosis
. N Engl J Med. 2014;370:33–40
7. Bellumkonda L, Subrahmanyan L, Jacoby D, Bonde P. Left ventricular assist device pump thrombosis
: Is there a role for glycoprotein IIb/IIIa inhibitors? ASAIO J. 2014;60:134–136
8. Cowger JA, Romano MA, Shah P, et al. Hemolysis
: A harbinger of adverse outcome after left ventricular assist device implant. J Heart Lung Transplant. 2014;33:35–43
9. Vercaemst L. Hemolysis
in cardiac surgery patients undergoing cardiopulmonary bypass: A review in search of a treatment algorithm. J Extra Corpor Technol. 2008;40:257–267
10. Marsden AL, Bazilevs Y, Long CC, Behr M. Recent advances in computational methodology for simulation of mechanical circulatory assist devices. Wiley Interdiscip Rev Syst Biol Med. 2014;6:169–188
11. Fraser KH, Taskin ME, Griffith BP, Wu ZJ. The use of computational fluid dynamics in the development of ventricular assist devices. Med Eng Phys. 2011;33:263–280
12. Medvitz RB Development and Validation of a Computational Fluid Dynamic Methodology for Pulsatile Blood Pump Design and Prediction of Thrombus Potential. PhD Thesis. 2008 Pennsylvania The Pennsylvania State University
13. Jamiolkowski MA, Woolley JR, Kameneva MV, Antaki JF, Wagner WR. Real time visualization and characterization of platelet deposition under flow onto clinically relevant opaque surfaces. J Biomed Mater Res A. 2015;103:1303–1311
14. Nanna JC, Navitsky MA, Topper SR, Deutsch S, Manning KB. A fluid dynamics study in a 50 cc pulsatile ventricular assist device: Influence of heart rate variability. J Biomech Eng. 2011;133:101002
15. Oley LA, Manning KB, Fontaine AA, Deutsch S. Off-design considerations of the 50cc Penn State Ventricular Assist Device. Artif Organs. 2005;29:378–386
16. Cooper BT, Roszelle BN, Long TC, Deutsch S, Manning KB. The influence of operational protocol on the fluid dynamics in the 12 cc Penn state pulsatile pediatric ventricular assist device: The effect of end-diastolic delay. Artif Organs. 2010;34:E122–E133
17. Shahraki ZH, Oscuii HN. Numerical investigation of three patterns of motion in an electromagnetic pulsatile VAD
. ASAIO J. 2014;60:304–310
18. Navitsky MA, Deutsch S, Manning KB. A thrombus susceptibility comparison of two pulsatile Penn State 50 cc left ventricular assist device designs. Ann Biomed Eng. 2013;41:4–16
19. Navitsky MA, Taylor JO, Smith AB, et al. Platelet adhesion to polyurethane urea under pulsatile flow conditions. Artif Organs. 2014;38:1046–1053
20. Shi Y, Brown AG, Lawford PV, Arndt A, Nuesser P, Hose DR. Computational modelling and evaluation of cardiovascular response under pulsatile impeller pump support. Interface Focus. 2011;1:320–337
21. Gohean JR, George MJ, Pate TD, Kurusz M, Longoria RG, Smalling RW. Verification of a computational cardiovascular system model comparing the hemodynamics of a continuous flow to a synchronous valveless pulsatile flow left ventricular assist device. ASAIO J. 2013;59:107–116
22. Taskin ME, Fraser KH, Zhang T, Wu C, Griffith BP, Wu ZJ. Evaluation of Eulerian and Lagrangian models for hemolysis
estimation. ASAIO J. 2012;58:363–372
23. Girdhar G, Xenos M, Alemu Y, et al. Device thrombogenicity emulation: A novel method for optimizing mechanical circulatory support device thromboresistance. PLoS One. 2012;7:e32463
24. Taskin ME, Zhang T, Fraser KH, Griffith BP, Wu ZJJ. Design optimization of a wearable artificial pump-lung device with computational modeling. J Med Devices. 2012;6:12
25. Topper SR, Navitsky MA, Medvitz RB, et al. The use of fluid mechanics to predict regions of microscopic thrombus formation in pulsatile VADs. Cardiovasc Eng Technol. 2014;5:54–69
26. Chang X, Gorbet M. The effect of shear on in vitro platelet and leukocyte material-induced activation. J Biomater Appl. 2013;28:407–415
27. Moosavi MH, Fatouraee N, Katoozian H. Finite element analysis of blood flow characteristics in a Ventricular Assist Device (VAD). Simul Model Pract Theory. 2009;17:654–663
28. Medvitz RB, Kreider JW, Manning KB, Fontaine AA, Deutsch S, Paterson EG. Development and validation of a computational fluid dynamics methodology for simulation of pulsatile left ventricular assist devices. ASAIO J. 2007;53:122–131
29. Lim KM, Kim IS, Choi SW, et al. Computational analysis of the effect of the type of LVAD flow on coronary perfusion and ventricular afterload. J Physiol Sci. 2009;59:307–316
30. Fraser KH, Zhang T, Taskin ME, Griffith BP, Wu ZJ. A quantitative comparison of mechanical blood damage parameters in rotary ventricular assist devices: Shear stress, exposure time and hemolysis
index. J Biomech Eng. 2012;134:081002
31. Corbett SC, Ajdari A, Coskun AU, Nayeb-Hashemi H. Effect of pulsatile blood flow on thrombosis
potential with a step wall transition. ASAIO J. 2010;56:290–295
32. Jarvis P, Tarbell JM, Frangos JA. An in vitro evaluation of an artificial heart. ASAIO Trans. 1991;37:27–32
33. Zhang Q, Gao B, Gu K, Chang Y, Xu J, Deuflhard P. The effect of varied support models of BJUT-IIVAD on coronary arterial blood flow and wall shear stress: A primary CFD study. ASAIO J. 2014;60:643–651
34. Jahren SE, Amacher R, Weber A, et al. Effects of Thoratec pulsatile ventricular assist device timing on the abdominal aortic wave intensity pattern. Am J Physiol Heart Circ Physiol. 2014;307:H1243–H1251