A homogenized two-phase computational framework for meso- and macroscale blood flow simulations.

BACKGROUND AND OBJECTIVE: Owing to the complexity of physics linked with blood flow and its associated phenomena, appropriate modeling of the multi-constituent rheology of blood is of primary importance. To this effect, various kinds of computational fluid dynamic models have been developed, each with merits and limitations. However, when additional physics like thrombosis and embolization is included within the framework of these models, computationally efficient scalable translation becomes very difficult. Therefore, this paper presents a homogenized two-phase blood flow framework with similar characteristics to a single fluid model but retains the flow resolution of a classical two-fluid model. The presented framework is validated against four different sets of experiments. METHODS: The two-phase model of blood presented here is based on the classical diffusion-flux framework. Diffusion flux models are known to be less computationally expensive than two-fluid multiphase models since the numerical implementation resembles single-phase flow models. Diffusion flux models typically use empirical slip velocity correlations to resolve the motion between phases. However, such correlations do not exist for blood. Therefore, a modified slip velocity equation is proposed, derived rigorously from the two-fluid governing equations. An additional drag law for red blood cells (RBCs) as a function of volume fraction is evaluated using a previously published cell-resolved solver. A new hematocrit-dependent expression for lift force on RBCs is proposed. The final governing equations are discretized and solved using the open-source software OpenFOAM. RESULTS: The framework is validated against four sets of experiments: (i) flow through a rectangular microchannel to validate RBC velocity profiles against experimental measurements and compare computed hematocrit distributions against previously reported simulation results (ii) flow through a sudden expansion microchannel for comparing experimentally obtained contours of hematocrit distributions and normalized cell-free region length obtained at different flowrates and inlet hematocrits, (iii) flow through two hyperbolic channels to evaluate model predictions of cell-free layer thickness, and (iv) flow through a microchannel that mimics crevices of a left ventricular assist device to predict hematocrit distributions observed experimentally. The simulation results exhibit good agreement with the results of all four experiments. CONCLUSION: The computational framework presented in this paper has the advantage of resolving the multiscale physics of blood flow while still leveraging numerical techniques used for solving single-phase flows. Therefore, it becomes an excellent candidate for addressing more complicated problems related to blood flow, such as modeling mechanical entrapment of RBCs within blood clots, predicting thrombus composition, and visualizing clot embolization.

Exploratory Simulation of Thrombosis in a Temporary LVAD Catheter Pump within a Virtual In-vivo Left Heart Environment.

Percutaneous catheter pumps are intraventricular temporary mechanical circulatory support (MCS) devices that are positioned across the aortic valve into the left ventricle (LV) and provide continuous antegrade blood flow from the LV into the ascending aorta (AA). MCS devices are most often computationally evaluated as isolated devices subject to idealized steady-state blood flow conditions. In clinical practice, MCS devices operate connected to or within diseased pulsatile native hearts and are often complicated by hemocompatibility related adverse events such as stroke, bleeding, and thrombosis. Whereas aspects of the human circulation are increasingly being simulated via computational methods, the precise interplay of pulsatile LV hemodynamics with MCS pump hemocompatibility remains mostly unknown and not well characterized. Technologies are rapidly converging such that next-generation MCS devices will soon be evaluated in virtual physiological environments that increasingly mimic clinical settings. The purpose of this brief communication is to report results and lessons learned from an exploratory CFD simulation of hemodynamics and thrombosis for a catheter pump situated within a virtual in-vivo left heart environment.

A fibrin enhanced thrombosis model for medical devices operating at low shear regimes or large surface areas.

Over the past decade, much of the development of computational models of device-related thrombosis has focused on platelet activity. While those models have been successful in predicting thrombus formation in medical devices operating at high shear rates (> 5000 s-1), they cannot be directly applied to low-shear devices, such as blood oxygenators and catheters, where emerging information suggest that fibrin formation is the predominant mechanism of clotting and platelet activity plays a secondary role. In the current work, we augment an existing platelet-based model of thrombosis with a partial model of the coagulation cascade that includes contact activation of factor XII and fibrin production. To calibrate the model, we simulate a backward-facing-step flow channel that has been extensively characterized in-vitro. Next, we perform blood perfusion experiments through a microfluidic chamber mimicking a hollow fiber membrane oxygenator and validate the model against these observations. The simulation results closely match the time evolution of the thrombus height and length in the backward-facing-step experiment. Application of the model to the microfluidic hollow fiber bundle chamber capture both gross features such as the increasing clotting trend towards the outlet of the chamber, as well as finer local features such as the structure of fibrin around individual hollow fibers. Our results are in line with recent findings that suggest fibrin production, through contact activation of factor XII, drives the thrombus formation in medical devices operating at low shear rates with large surface area to volume ratios.

Creating High-Resolution Microscopic Cross-Section Images of Hardwood Species Using Generative Adversarial Networks.

Microscopic wood identification plays a critical role in many economically important areas in wood science. Historically, producing and curating relevant and representative microscopic cross-section images of wood species is limited to highly experienced and trained anatomists. This manuscript demonstrates the feasibility of generating synthetic microscopic cross-sections of hardwood species. We leveraged a publicly available dataset of 119 hardwood species to train a style-based generative adversarial network (GAN). The proposed GAN generated anatomically accurate cross-section images with remarkable fidelity to actual data. Quantitative metrics corroborated the capacity of the generative model in capturing complex wood structure by resulting in a Fréchet inception distance score of 17.38. Image diversity was calculated using the Structural Similarity Index Measure (SSIM). The SSIM results confirmed that the GAN approach can successfully synthesize diverse images. To confirm the usefulness and realism of the GAN generated images, eight professional wood anatomists in two experience levels participated in a visual Turing test and correctly identified fake and actual images at rates of 48.3 and 43.7%, respectively, with no statistical difference when compared to random guess. The generative model can synthesize realistic, diverse, and meaningful high-resolution microscope cross-section images that are virtually indistinguishable from real images. Furthermore, the framework presented may be suitable for improving current deep learning models, helping understand potential breeding between species, and may be used as an educational tool.

Month-long Respiratory Support by a Wearable Pumping Artificial Lung in an Ovine Model.

BACKGROUND: A wearable artificial lung could improve lung transplantation outcomes by easing implementation of physical rehabilitation during long-term pretransplant respiratory support. The Modular Extracorporeal Lung Assist System (ModELAS) is a compact pumping artificial lung currently under development. This study evaluated the long-term in vivo performance of the ModELAS during venovenous support in awake sheep. Feedback from early trials and computational fluid dynamic analysis guided device design optimization along the way. METHODS: The ModELAS was connected to healthy sheep via a dual-lumen cannula in the jugular vein. Sheep were housed in a fixed-tether pen while wearing the device in a holster during support. Targeted blood flow rate and support duration were 2-2.5 L/min and 28-30 days, respectively. Anticoagulation was maintained via systemic heparin. Device pumping and gas exchange performance and hematologic indicators of sheep physiology were measured throughout support. RESULTS: Computational fluid dynamic-guided design modifications successfully decreased pump thrombogenicity from initial designs. For the optimized design, 4 of 5 trials advancing past early perioperative and cannula-related complications lasted the full month of support. Blood flow rate and CO2 removal in these trials were 2.1 ± 0.3 L/min and 139 ± 15 mL/min, respectively, and were stable during support. One trial ended after 22 days of support due to intradevice thrombosis. Support was well tolerated by the sheep with no signs of hemolysis or device-related organ impairment. CONCLUSIONS: These results demonstrate the ability of the ModELAS to provide safe month-long support without consistent deterioration of pumping or gas exchange capabilities.

Blood Recirculation Enhances Oxygenation Efficiency of Artificial Lungs.

Ambulating patients on extracorporeal membrane oxygenation (ECMO) or extracorporeal CO2 removal (ECCO2R) improves outcomes. These systems would further simplify ambulation if made more compact. This study investigates blood recirculation to decrease device size by increasing efficiency. The required hollow fiber membrane (HFM) area was determined by numerically modeling gas transfer. An oxygenation device with recirculating blood flow was designed using computational fluid dynamics (CFD). Hydrodynamic performance and shear stresses of the device were analyzed using CFD at 2,000, 2,250 and 2,500 RPM. A prototype (0.38 m) was manufactured for in-vitro oxygenation testing. Oxygenation was measured at a constant 3.5 L/min blood flow while recirculation flow rate varied up to 6.5 L/min. Hemolysis was measured at 3.5 L/min blood flow and 6.5 L/min recirculation flow. A 0.3 m prototype device was used to test in-vitro ECCO2R recirculation at a constant 500 ml/min blood flow rate and recirculation flow rates up to 5.5 L/min. Computational fluid dynamics analysis showed that the oxygenation device could produce over 250 mm Hg while maintaining 3.5 L/min blood flow and 6.5 L/min recirculation flow. The model predicted oxygenation within 8% and overestimated ECCO2R by up to 32%. Measured gas transfer was 180 ml O2/min and 62 ml CO2/min. Normalized index of hemolysis contribution of the HFM was 0.012 gm/100 L.

In Vivo 5 Day Animal Studies of a Compact, Wearable Pumping Artificial Lung.

Recent studies show improved outcomes in ambulated lung failure patients. Ambulation still remains a challenge in these patients. This necessitates development of more compact and less cumbersome respiratory support specifically designed to be wearable. The Paracorporeal Ambulatory Assist Lung (PAAL) is being designed for providing ambulatory support in lung failure patients during bridge to transplant or recovery. We previously published in vitro and acute in vivo results of the PAAL. This study further evaluates the PAAL for 5 days. Five-day in vivo studies with the PAAL were conducted in 50-60 kg sheep after heparinization (activated clotting time range: 190-250 s) and cannulation with a 27 Fr. Avalon Elite dual-lumen cannula. The animals were able to move freely in a stanchion while device flow, resistance, and hemodynamics were recorded hourly. Oxygenation and hemolysis were measured daily. Platelet activation, blood chemistry, and comprehensive blood counts are reported for preoperatively, on POD 0, and POD 5. Three animals survived for 5 days. No study termination resulted from device failure. One animal was terminated on POD 0 and one animal was terminated at POD 3. The device was operated between 1.93 and 2.15 L/min. Blood left the device 100% oxygenated. Plasma-free hemoglobin ranged 10.8-14.5 mg/dl. CD62-P expression was under 10%. Minimal thrombus was seen in devices at explant. Chronic use of the PAAL in awake sheep is promising based on our study. There were no device-related complications over the study course. This study represents the next step in our pathway to eventual clinical translation.

Bench Validation of a Compact Low-Flow CO2 Removal Device.

BACKGROUND: There is increasing evidence demonstrating the value of partial extracorporeal CO2 removal (ECCO2R) for the treatment of hypercapnia in patients with acute exacerbations of chronic obstructive pulmonary disease and acute respiratory distress syndrome. Mechanical ventilation has traditionally been used to treat hypercapnia in these patients, however, it has been well-established that aggressive ventilator settings can lead to ventilator-induced lung injury. ECCO2R removes CO2 independently of the lungs and has been used to permit lung protective ventilation to prevent ventilator-induced lung injury, prevent intubation, and aid in ventilator weaning. The Low-Flow Pittsburgh Ambulatory Lung (LF-PAL) is a low-flow ECCO2R device that integrates the fiber bundle (0.65 m2) and centrifugal pump into a compact unit to permit patient ambulation. METHODS: A blood analog was used to evaluate the performance of the pump at various impeller rotation rates. In vitro CO2 removal tested under normocapnic conditions and 6-h hemolysis testing were completed using bovine blood. Computational fluid dynamics and a mass-transfer model were also used to evaluate the performance of the LF-PAL. RESULTS: The integrated pump was able to generate flows up to 700 mL/min against the Hemolung 15.5 Fr dual lumen catheter. The maximum vCO2 of 105 mL/min was achieved at a blood flow rate of 700 mL/min. The therapeutic index of hemolysis was 0.080 g/(100 min). The normalized index of hemolysis was 0.158 g/(100 L). CONCLUSIONS: The LF-PAL met pumping, CO2 removal, and hemolysis design targets and has the potential to enable ambulation while on ECCO2R.

In Vitro Characterization of the Pittsburgh Pediatric Ambulatory Lung.

Acute and chronic respiratory failure are a significant source of pediatric morbidity and mortality. Current respiratory support options used to bridge children to lung recovery or transplantation typically render them bedridden and can worsen long-term patient outcomes. The Pittsburgh Pediatric Ambulatory Lung (P-PAL) is a wearable pediatric blood pump and oxygenator (0.3 m surface area) integrated into a single compact unit that enables patient ambulation. The P-PAL is intended for long-term use and designed to provide up to 90% of respiratory support in children weighing 5-25 kg. Computational fluid dynamics and numerical gas exchange modeling were used to design the P-PAL and predict its performance. A P-PAL prototype was then used to obtain pressure versus flow curves at various impeller rotation rates using a blood analog fluid. In vitro oxygen exchange rates were obtained in blood in accordance with ISO standard 7199. The normalized index of hemolysis (NIH) was measured over a 6 hour period at blood flow rates of 1 and 2.5 L/min. The P-PAL provided blood flows of 1-2.5 L/min against the pressure drop associated with its intended-use pediatric cannulas. The oxygen exchange rate reached a maximum of 108 ml/min at a blood flow rate of 2.5 L/min and met our respiratory support design target. Device-induced hemolysis was low with NIH values of 0.022-0.027 g/100 L in the intended blood flow rate range. In conclusion, the current P-PAL design met our pumping, oxygenation, and hemolysis specifications and has the potential to improve treatment for pediatric respiratory failure.

Platelet activation and erythrocyte lysis during brief exposure of blood to pathophysiological shear stress in vitro.

BACKGROUND: Interaction of von Willebrand factor (VWF) with circulating platelets is the trigger for thrombosis in a region of arterial stenosis. These events are typically studied in vitro under conditions where platelets adhere to a VWF-coated surface. Our approach assesses platelet responses in the absence of adhesion. OBJECTIVE: To characterize extent of platelet activation and erythrocyte lysis in an artificial stenosis model. METHODS: Whole blood is perfused through a length of polyetheretherketone tubing that includes an artificial stenosis, comprising narrow-bore (89-381 µm) tubing. Secretion of [14C] serotonin and hemoglobin release was measured to evaluate platelet activation and hemolysis respectively at various perfusion rates and different stenosis dimensions. RESULTS: Platelet activation and erythrocyte lysis increased progressively with increasing perfusion rate and decreasing stenosis diameter; the length of the stenosis had negligible influence. Modest platelet activation (5-10% secretion of [14C] serotonin) occurred without significant erythrocyte lysis under a limited range of perfusion conditions (4-6 mL/min flow through a 127 µm stenosis). CONCLUSIONS: Our experimental approach mimics conditions in severe arterial stenosis or a mechanical heart valve. It could be a valuable aid in the development of novel drugs to treat arterial thrombosis and in the design of heart valves.

In vitro and in vivo evaluation of a novel integrated wearable artificial lung.

BACKGROUND: Conventional extracorporeal membrane oxygenation (ECMO) is cumbersome and is associated with high morbidity and mortality. We are currently developing the Pittsburgh Ambulatory Assist Lung (PAAL), which is designed to allow for ambulation of lung failure patients during bridge to transplant or recovery. In this study, we investigated the in vitro and acute in vivo performance of the PAAL. METHODS: The PAAL features a 1.75-inch-diameter, cylindrical, hollow-fiber membrane (HFM) bundle of stacked sheets, with a surface area of 0.65 m2 integrated with a centrifugal pump. The PAAL was tested on the bench for hydrodynamic performance, gas exchange and hemolysis. It was then tested in 40- to 60-kg adult sheep (n = 4) for 6 hours. The animals were cannulated with an Avalon Elite 27Fr dual-lumen catheter (DLC) inserted through the right external jugular into the superior vena cava (SVC), right atrium (RA) and inferior vena cava (IVC). RESULTS: The PAAL pumped >250 mm Hg at 3.5 liters/min at a rotation speed of 2,100 rpm. Oxygenation performance met the target of 180 ml/min at 3.5 liters/min of blood flow in vitro, resulting in a gas-exchange efficiency of 278 ml/min/m2. The normalized index of hemolysis (NIH) for the PAAL and cannula was 0.054 g per 100 liters (n = 2) at 3.5 liters/min, as compared with 0.020 g per 100 liters (n = 2) for controls (DLC cannula and a Centrimag pump). Plasma-free hemoglobin (pfHb) was <20 mg/dl for all animals. Blood left the device 100% oxygenated in vivo and oxygenation reached 181 ml/min at 3.8 liters/min. CONCLUSION: The PAAL met in vitro and acute in vivo performance targets. Five-day chronic sheep studies are planned for the near future.

Effect of impeller design and spacing on gas exchange in a percutaneous respiratory assist catheter.

Providing partial respiratory assistance by removing carbon dioxide (CO2 ) can improve clinical outcomes in patients suffering from acute exacerbations of chronic obstructive pulmonary disease and acute respiratory distress syndrome. An intravenous respiratory assist device with a small (25 Fr) insertion diameter eliminates the complexity and potential complications associated with external blood circuitry and can be inserted by nonspecialized surgeons. The impeller percutaneous respiratory assist catheter (IPRAC) is a highly efficient CO2 removal device for percutaneous insertion to the vena cava via the right jugular or right femoral vein that utilizes an array of impellers rotating within a hollow-fiber membrane bundle to enhance gas exchange. The objective of this study was to evaluate the effects of new impeller designs and impeller spacing on gas exchange in the IPRAC using computational fluid dynamics (CFD) and in vitro deionized water gas exchange testing. A CFD gas exchange and flow model was developed to guide a progressive impeller design process. Six impeller blade geometries were designed and tested in vitro in an IPRAC device with 2- or 10-mm axial spacing and varying numbers of blades (2-5). The maximum CO2 removal efficiency (exchange per unit surface area) achieved was 573 ± 8 mL/min/m(2) (40.1 mL/min absolute). The gas exchange rate was found to be largely independent of blade design and number of blades for the impellers tested but increased significantly (5-10%) with reduced axial spacing allowing for additional shaft impellers (23 vs. 14). CFD gas exchange predictions were within 2-13% of experimental values and accurately predicted the relative improvement with impellers at 2- versus 10-mm axial spacing. The ability of CFD simulation to accurately forecast the effects of influential design parameters suggests it can be used to identify impeller traits that profoundly affect facilitated gas exchange.

Efficient, physiologically realistic lung airflow simulations.

One of the key challenges for computational fluid dynamics (CFD) simulations of human lung airflow is the sheer size and complexity of the complete, multiscale geometry of the bronchopulmonary tree. Since 3-D CFD simulations of the full airway tree are currently intractable, researchers have proposed reduced geometry models in which multiple airway paths are truncated downstream of the first few generations. This paper investigates a recently proposed method for closing the CFD model by application of physiologically correct boundary conditions at truncated outlets. A realistic, reduced geometry model of the lung airway based on CT data has been constructed up to generation 18, including extrathoracic, bronchi, and bronchiole regions. Results indicate that the new method yields reasonable results for pressure drop through the airway, at a small fraction of the cost of fully resolved simulations.

Multilaboratory particle image velocimetry analysis of the FDA benchmark nozzle model to support validation of computational fluid dynamics simulations.

This study is part of a FDA-sponsored project to evaluate the use and limitations of computational fluid dynamics (CFD) in assessing blood flow parameters related to medical device safety. In an interlaboratory study, fluid velocities and pressures were measured in a nozzle model to provide experimental validation for a companion round-robin CFD study. The simple benchmark nozzle model, which mimicked the flow fields in several medical devices, consisted of a gradual flow constriction, a narrow throat region, and a sudden expansion region where a fluid jet exited the center of the nozzle with recirculation zones near the model walls. Measurements of mean velocity and turbulent flow quantities were made in the benchmark device at three independent laboratories using particle image velocimetry (PIV). Flow measurements were performed over a range of nozzle throat Reynolds numbers (Re(throat)) from 500 to 6500, covering the laminar, transitional, and turbulent flow regimes. A standard operating procedure was developed for performing experiments under controlled temperature and flow conditions and for minimizing systematic errors during PIV image acquisition and processing. For laminar (Re(throat)=500) and turbulent flow conditions (Re(throat)≥3500), the velocities measured by the three laboratories were similar with an interlaboratory uncertainty of ∼10% at most of the locations. However, for the transitional flow case (Re(throat)=2000), the uncertainty in the size and the velocity of the jet at the nozzle exit increased to ∼60% and was very sensitive to the flow conditions. An error analysis showed that by minimizing the variability in the experimental parameters such as flow rate and fluid viscosity to less than 5% and by matching the inlet turbulence level between the laboratories, the uncertainties in the velocities of the transitional flow case could be reduced to ∼15%. The experimental procedure and flow results from this interlaboratory study (available at http://fdacfd.nci.nih.gov) will be useful in validating CFD simulations of the benchmark nozzle model and in performing PIV studies on other medical device models.

DigitalLung: application of high-performance computing to biological system simulation.

The DigitalLung project represents an attempt to develop a multi-scale capability for simulating human respiration with application to predicting the effects of inhaled particulate matter. To accomplish this objective, DigitalLung integrates macroscale models of integrative human physiology, meso-to-microscale computational fluid dynamics simulations of a breathing human lung, meso-to-nanoscale particle transport and deposition models, and micro-to-nanoscale physical and chemical characterizations of particulate and their mass transfer through the mucosal layer to the epithelium. This chapter describes preliminary results and areas of ongoing research.

Effects of turbulent stresses upon mechanical hemolysis: experimental and computational analysis.

Experimental and computational studies were performed to elucidate the role of turbulent stresses in mechanical blood damage (hemolysis). A suspension of bovine red blood cells (RBC) was driven through a closed circulating loop by a centrifugal pump. A small capillary tube (inner diameter 1 mm and length 70 mm) was incorporated into the circulating loop via tapered connectors. The suspension of RBCs was diluted with saline to achieve an asymptotic apparent viscosity of 2.0 +/- 0.1 cP at 23 degrees C to produce turbulent flow at nominal flow rate and pressure. To study laminar flow at the identical wall shear stresses in the same capillary tube, the apparent viscosity of the RBC suspension was increased to 6.3 +/- 0.1 cP (at 23 degrees C) by addition of Dextran-40. Using various combinations of driving pressure and Dextran mediated adjustments in dynamic viscosity Reynolds numbers ranging from 300-5,000 were generated, and rates of hemolysis were measured. Pilot studies were performed to verify that the suspension media did not affect mechanical fragility of the RBCs. The results of these bench studies demonstrated that, at the same wall shear stress in a capillary tube, the level of hemolysis was significantly greater (p < 0.05) for turbulent flow as compared with laminar flow. This confirmed that turbulent stresses contribute strongly to blood mechanical trauma. Numerical predictions of hemolysis obtained by computational fluid dynamic modeling were in good agreement with these experimental data.

Computational fluid dynamics analysis of a maglev centrifugal left ventricular assist device.

The fluid dynamics of the Thoratec HeartMate III (Thoratec Corp., Pleasanton, CA, U.S.A.) left ventricular assist device are analyzed over a range of physiological operating conditions. The HeartMate III is a centrifugal flow pump with a magnetically suspended rotor. The complete pump was analyzed using computational fluid dynamics (CFD) analysis and experimental particle imaging flow visualization (PIFV). A comparison of CFD predictions to experimental imaging shows good agreement. Both CFD and experimental PIFV confirmed well-behaved flow fields in the main components of the HeartMate III pump: inlet, volute, and outlet. The HeartMate III is shown to exhibit clean flow features and good surface washing across its entire operating range.

Predicting membrane oxygenator pressure drop using computational fluid dynamics.

Three-dimensional computational fluid dynamic (CFD) simulations of membrane oxygenators should allow prediction of spatially dependent variables and subsequent shape optimization. Fiber bed complexity and current computational limitations require the use of approximate models to predict fiber drag effects in complete device simulations. A membrane oxygenator was modified to allow pressure measurement along the fiber bundle in all cardinal axes. Experimental pressure drop information with water perfusion was used to calculate the permeability of the fiber bundle. A three-dimensional CFD model of a commercial membrane oxygenator was developed to predict pressure drops throughout the device. Darcy's Law was used to account for the viscous drag of the fibers and was incorporated as a momentum loss term in the conservation equations. Close agreement was shown between experimental and simulated pressure drops at lower flow rates, but the simulated pressure drops were lower than experimental results at higher flows. Alternate models of fiber drag effects and flow field visualization are suggested as means to potentially improve the accuracy of the flow simulation. Computational techniques coupled with experimental verification offer insight into model validity and show promise for the development of accurate three-dimensional simulations of membrane oxygenators.