Acute and chronic respiratory failure represent a significant source of morbidity and mortality within the pediatric population. Although acute lung injury (ALI) is less prevalent in children than adults, ALI accounts for up to 30% of deaths within some pediatric intensive care units.1 The incidence of ALI is also suspected to be underestimated in the literature due to alternative diagnoses.2 In regard to chronic lung disease, lung and lung–heart transplants have become increasingly more frequent for children with end-stage cystic fibrosis, pulmonary hypertension, and pulmonary fibrosis.3 The International Society for Heart and Lung Transplantation Registry reports that from 1986 to 2013, 2,091 lung and 689 heart–lung transplants were performed worldwide in pediatric patients with these numbers continuing to rise in recent years.4 This datum also likely underestimates the true number of transplants due to those procedures not reported to the Registry. The duration that pediatric patients are on a waiting list before receiving a donor lung can be less than a month to a matter of years. Between 2009 and 2011, approximately 80% of patients waited more than 1 month with the majority of those children receiving a transplant in 1–6 months.5
Mechanical ventilation (MV) and extracorporeal membrane oxygenation (ECMO) are currently used to support children with lung failure as a bridge to lung recovery or transplantation. Multiple studies, however, have demonstrated poor post-transplant outcomes related to the use of these procedures on children awaiting lung transplantation.6,7 Consequently, many transplant centers consider the pre-transplant use of MV and ECMO a contraindication to lung transplantation.3,7 Long-term use of MV can result in barotrauma and volutrauma.8 The successful use of conventional ECMO can be hindered by its numerous system components and generally cumbersome nature. In addition, both treatments typically require immobilization of the patient, which often leads to a progressive deconditioning that can increase post-transplant morbidity and mortality.9,10 Ambulation and active rehabilitation during the pre-transplant period, however, have been shown to counteract such deconditioning and improve patient outcomes.11,12 Thus, there is a clear unmet need for development of a more efficient and smaller pediatric respiratory assist device that allows patient ambulation.
Here we describe the initial development of the Pittsburgh Pediatric Ambulatory Lung (P-PAL). The P-PAL integrates a centrifugal pump with a highly efficient oxygenating hollow fiber membrane (HFM) bundle into a single compact unit that is meant to be wearable and allow for ambulation. The P-PAL is proposed for longer-term use (1–3 months without device exchange) in children weighing 5–25 kg and intended to replace MV or conventional ECMO as a bridge to transplant or recovery during lung failure. During use of the P-PAL, blood flow is redirected from the patient to the P-PAL via a venous cannula, driven through the gas-exchanging HFM bundle by the integrated pump, and then returned to the patient via an arterial cannula. Primary performance specifications were to provide up to 90% of respiratory support while pumping blood 1–2.5 L/min and maintaining low device-induced hemolysis. Minimization of the HFM surface area and overall device size were also a primary focus to both minimize blood-contacting surfaces and allow for patient mobility. Computational fluid dynamics (CFD) and numerical gas exchange modeling were used to design the P-PAL as well as to analyze its performance. A P-PAL prototype was fabricated and used to characterize in vitro pumping and oxygenation capabilities over the intended operating conditions. Six hour hemolysis evaluations were also conducted to assess the level of blood damage induced by the device.
Materials and Methods
P-PAL Design and Fabrication
A schematic and picture of the P-PAL prototype are shown in Figure 1. The centrifugal pump utilizes an enclosed impeller to draw blood in through the venous cannula and drive flow through the HFM bundle and arterial cannula. The P-PAL housing and impeller were modeled in SolidWorks CAD software (Dassault Systèmes, SolidWorks Corp., Waltham, MA) and machined from clear acrylic and polycarbonate, respectively. After machining, the housing surfaces were polished to a translucent, glass-like finish, whereas the impeller surfaces were vapor polished. Permanent magnets were embedded in the impeller to enable magnetic coupling with an external motor that drives impeller rotation. The two-pivot bearing system is composed of ceramic pivots at either end of the impeller shaft that are supported by ultra-high molecular weight polyethylene cups. Similar to an adult pump-oxygenator currently under development by our group,13,14 the P-PAL employs a cylindrical, stacked-type HFM bundle with a minimized, circular cross-sectional area (1.75 inch bundle diameter). The bundle has a porosity of 0.5 and total membrane surface area available for gas exchange of 0.3 m2. The bundle was manufactured from commercially available polymethylpentene hollow fiber sheets (OXYPLUS, 44 fibers per inch, Membrana GmbH, Wuppertal, Germany) using previously described methods.13 Briefly, circular fiber sheets were stacked alternatingly at a 14° crossing angle and potted using a polyurethane adhesive (Vertellus Performance Materials Inc., Greensboro, NC). The current P-PAL prototype has a total weight of approximately 2.5 lb, but this weight will be significantly reduced once the P-PAL design is validated and fabrication moves to injection molded components.
Oxygen transfer in the P-PAL was modeled using a previously developed mass transfer correlation.15 The mass balance for oxygen transfer in a cylindrical HFM bundle is given by Equation 1.
where is the volumetric blood flow rate through the bundle (ml/s), is the concentration of oxygen in the blood (ml O2/ml blood), is the bundle radius (cm), is the mass transfer coefficient [ml O2/(cm2·s·mm Hg)], is the surface area-to-volume ratio of the bundle (cm2/cm3), is the oxygen partial pressure difference between the blood and gas sides of the fiber (mm Hg), and is the axial coordinate (cm). accounts for both dissolved and chemically bound oxygen in the blood and is given by Equation 2.
where is the solubility of oxygen in blood [3E-5 ml O2/(ml blood·mm Hg)], is the partial pressure of oxygen in the blood (mm Hg), is the binding capacity of saturated blood (0.167 ml O2/ml blood), and is the fractional blood oxygen saturation. Substituting Equation 2 into Equation 1 results in a first order differential equation describing the change in oxygen partial pressure as a function of (Equation 3).
where is the slope of the oxyhemoglobin dissociation curve. The mass transfer coefficient, , is determined by Equation 4.
is the Sherwood number defined as , where is the diffusivity of oxygen in blood (1.8E-5 cm2/s) and is the fiber diameter (cm). is the Reynolds number defined as , where is the superficial blood velocity (cm/s), is the hydraulic diameter characteristic of fiber bed packing (cm),16 and is the kinematic blood viscosity (0.023 cm2/s). is the Schmidt number defined as , where is a previously used15 effective diffusivity for oxygen in blood (cm2/s) and accounts for chemically bound oxygen. The differential equation resulting from Equations 1–4 was solved in MATLAB (MathWorks, Natick, MA) using the built-in ordinary differential equation solver based on the Runge–Kutta method. Oxygen transfer rates were calculated for the P-PAL at blood flow rates of 1–2.5 L/min to determine the required HFM surface area to meet oxygenation requirements.
Blood flow velocities and pressures within the P-PAL were modeled via laminar CFD analysis performed using ANSYS Fluent v17 (ANSYS, Canonsburg, PA). Blood was treated as a homogeneous incompressible fluid of density 1050 kg/m3 and constant viscosity of 3.5 cP. The fiber bundle was modeled as porous media with a uniform viscous resistance17 of 1E9 1/m2 and a fluid porosity of 0.5. The CFD mesh consisted of 4M tetrahedral cells, and impeller motion was implemented using a frozen relative motion frame of reference.
In vitro pump performance of the P-PAL was evaluated using a carboxymethyl cellulose (CMC) sodium salt (Sigma Aldrich, St. Louis, MO) solution (8.5 g/L) as the working fluid. The CMC solution was prepared to have a viscosity of 3.5 cP when heated to 37°C and the viscosity verified using a capillary viscometer (Cannon Instrument Company, State College, PA). The experimental circuit consisted of the P-PAL connected to an 800 ml compliant blood reservoir (Medtronic, Minneapolis, MN). The reservoir was submerged within a heated water bath during testing to maintain the temperature of the working fluid at 37 ± 2°C. Pressures at the inlet and outlet of the P-PAL were measured in triplicate using a Honeywell 143 PC03D transducer (Honeywell International Inc., Morris Plains, NJ) at flow rates of 0–3 L/min and impeller rotation rates of 1000, 1200, 1400, 1600, and 1800 RPM. Flow rate was measured using an ultrasonic flow probe (Transonic Systems Inc., Ithaca, NY) and varied at each rotation rate using a Hoffman clamp distal to the device.
During clinical use the, P-PAL is intended to be used with a 14 Fr DLP straight tip arterial cannula (#75014, Medtronic, Minneapolis, MN) and 18 Fr DLP malleable venous cannula (#68118, Medtronic, Minneapolis, MN). The pressure drop of those cannulas was measured using methods similar to those described above. Measurements were made at flow rates of 0–3 L/min with flow being driven by a Biomedicus BP 80-X pump (Medtronic, Minneapolis, MN).
In Vitro Oxygenation
In vitro oxygen exchange rates were measured in accordance with ISO standard 7199 using a previously described experimental circuit13 and bovine or porcine blood collected from a local slaughterhouse. Before use, blood was filtered (40 µm filter, Pall Biomedical Inc., Fajardo, PR) and treated with heparin (15 IU/ml) and gentamicin (0.1 mg/ml). Blood was first preconditioned to venous conditions (O2 saturation = 65% ± 5%, pCO2 = 45 ± 5 mm Hg) via recirculation through a Medtronic Affinity NT oxygenator (Medtronic, Minneapolis, MN). Once venous blood conditions were achieved, sweep gas to the Affinity oxygenator was discontinued and tubing was clamped to produce single-pass blood flow through the P-PAL for oxygen exchange rate measurements. Blood temperature was maintained at 37 ± 2°C throughout the experiment via a Polyscience 210 heater (Polyscience Inc., Niles, IL) connected to the heat exchanger of the Affinity oxygenator. P-PAL oxygen exchange rates were evaluated at blood flow rates from 1 to 2.5 L/min with impeller rotation rates of 1100–1700 RPM to match operating conditions predicted from pump testing. A Hoffman clamp immediately distal to the P-PAL outlet was used to achieve specified flow rates, which were measured via an ultrasonic flow probe. Pure oxygen was used as the sweep gas and maintained at a constant gas flow rate of 6 L/min using a GR series mass flow controller (Fathom Technologies, Georgetown, TX). Blood samples were taken at the inlet and outlet of the test device and analyzed using a RAPIDPoint 405 blood gas analyzer with co-oximetry (Siemens Healthcare Diagnostics Inc., Tarrytown, NY). Six repeated measurements were taken for each condition over the course of two independent experiments. Oxygen exchange rates were calculated from inlet and outlet oxygen partial pressures and saturations using Equation 5.
where is the oxygen exchange rate (ml/min), is the blood flow rate (L/min), is the solubility of oxygen in blood [3E-2 ml O2/(L blood·mm Hg)], is the oxygen partial pressure difference across the device (mm Hg), is the hemoglobin binding capacity (1.34 ml O2/g), is the hemoglobin concentration (g/dl), and is the fractional oxygen saturation difference across the device.
In Vitro Hemolysis
Blood damage was evaluated at flow rates of 1 and 2.5 L/min in accordance with ASTM F1841-97 using bovine blood collected from a local slaughterhouse. After collection, blood was filtered and treated as previously specified for oxygenation testing. The P-PAL was evaluated in parallel with a control circuit utilizing a Lilliput 2 oxygenator (Sorin Group, Mirandola, Italy) and either a PediMag (used at 1 L/min) or CentriMag (used at 2.5 L/min) blood pump (Thoratec, Pleasanton, CA). Evaluation was performed using a continuous flow circuit consisting of the test device (P-PAL or control) connected to an 800 ml compliant blood reservoir. The compliant reservoir was submerged within a heated water bath during testing to maintain a blood temperature of 37 ± 2°C. A Hoffman clamp distal to the device was used to adjust afterload to achieve a pressure differential of 100 mm Hg across the device. For each operating condition evaluated, blood (hematocrit = 30%) was circulated for a period of 6 hours during which blood samples were collected every 30 minutes. Hematocrit and plasma free hemoglobin measurements were taken for each sample and used to calculate a normalized index of hemolysis (NIH). Details regarding measurement techniques and NIH calculation have been previously published.18 Three independent trials were conducted for each condition evaluated.
SPSS 22 (IBM Corporation, Armonk, NY) was used to perform statistical analysis. A two-way ANOVA was used to determine the significance of the effects of device type and flow rate on NIH. All error bars in presented figures represent standard deviations.
Typical CFD results shown in Figure 2 demonstrate adequate pressure generation and uniform flow distribution throughout the bundle. Generated pressure heads predicted by CFD were 88 mm Hg for 1100 RPM, 1 L/min and 196 mm Hg for 1700 RPM, 2.5 L/min. The maximum shear stress in the device for these operating conditions was located on the impeller surfaces and was less than 200 Pa. Flow through the impeller channels was shown to have uniform distribution and exhibited no flow separation. For both operating conditions evaluated, the device was remarkably free of flow stasis and deleterious flow recirculation.
The pressure head generated by the P-PAL as a function of flow rate for impeller rotation rates varying from 1000 to 1800 RPM is shown in Figure 3. Figure 3 also shows the pressure requirements for operating the P-PAL. The pressure requirements are the sum of the pressure drop of the intended-use cannulas and the physiologic pressure differential of the points of cannulation. Right atrium (anticipated position of venous cannula) and pulmonary artery (anticipated position of arterial cannula) pressures of 0 and 50 mm Hg, respectively, were assumed. The P-PAL achieved the targeted flow rates of 1–2.5 L/min at impeller rotation rates between 1100 and 1700 RPM.
In Vitro and Modeled Oxygenation
Measured in vitro oxygen exchange rates as well as those predicted by the model are shown in Figure 4. Measured oxygen exchange increased with increasing blood flow rate and reached a maximum of 108 ml/min at a blood flow rate of 2.5 L/min. At 2.5 L/min the P-PAL met the oxygenation design target of 106 ml/min, which is approximately 90% of the oxygenation needs for a 25 kg child (based off of body surface area estimated via the Mosteller formula19).
In Vitro Hemolysis
The increase in plasma free hemoglobin over the 6 hour period of measurement exhibited a strong linear trend for all conditions with an average R2 value of 0.91 among all trials (results not shown). The calculated NIH values for the P-PAL and control circuit are shown in Figure 5. The NIH was similar between the P-PAL and control circuit with statistical analysis showing no significant effect of device type on NIH.
Current treatment options for pediatric lung failure are limited and may contribute to patient morbidity and mortality via their need for patient immobilization.9,10 To address this need, our group has begun development of the P-PAL. The P-PAL is a compact, wearable pediatric artificial lung designed to provide long-term respiratory support while enabling patient mobility. A functional P-PAL prototype was fabricated and used to perform a comprehensive benchtop evaluation. In vitro testing of the P-PAL demonstrated that the device is able to meet all specified performance targets.
MV and conventional ECMO are currently used to bridge children with lung failure to recovery or transplantation, but neither option is ideal for long-term support. Such treatments typically leave children bedridden and can often increase morbidity and mortality.9 Physical therapy has been shown to be beneficial for patients on long-term respiratory support,11,12 but no current clinical devices readily allow for ambulation. Once implemented clinically, the P-PAL will have the potential to improve treatment for children on respiratory support.
The P-PAL achieved the required pressure head generation predicted for in vivo conditions at flow rates up to 2.5 L/min. The physiologic pressures assumed in determining the required pressure head represent estimates for children with severe pulmonary hypertension. Thus, the P-PAL is capable of providing the necessary blood flow rates in even the most critically ill pediatric lung failure patients. Pressure heads predicted from CFD correlated well with in vitro testing with a maximum percent error of 2%. Based on flow uniformity from CFD modeling, good surface washing characteristics (i.e., low thrombogenicity potentials) are predicted throughout the device for both operating conditions evaluated.
Oxygenation performance is considered to be the most important gas exchange parameter for the P-PAL. In vivo studies of oxygenators with similar intended blood flow rates have shown the need to introduce CO2 into the sweep gas to limit its removal and avoid hypocapnia.14,20,21 Thus, oxygen transfer rate was the focus of gas exchange evaluation in this study. The P-PAL met targeted oxygenation specifications at the appropriate blood flow rates. The estimated cardiac output of the intended patient population ranges from 1.3 L/min (5 kg child) to 4.3 L/min (25 kg child).22 Estimated oxygen uptake for the same patients ranges from 35 ml/min (5 kg child) to 118 ml/min (25 kg child).19 Therefore, the targeted 90% of respiratory support is approximately 32 and 106 ml/min for a 5 kg and 25 kg child, respectively. Oxygenation results have demonstrated that the P-PAL is able to achieve these oxygen exchange rates at blood flow rates below the corresponding estimated cardiac outputs. Measured oxygen exchange rates were in good agreement with those predicted from numerical modeling with a maximum percent of error of 9%.
The P-PAL also maintained a low degree of blood damage over the course of 6 hours of continuous operation. P-PAL NIH values were not statistically different from those for the control circuit, which utilized a comparable nonintegrated pump/oxygenator combination currently in clinical use. Additionally, P-PAL NIH values were approximately 30–65% lower than those previously published for the only similar device in development, the PediPL,23 over the intended blood flow rates. The maximum shear stress predicted by CFD was less than that associated with significant hemolysis24 and thus correlated well with the low in vitro hemolysis measurements.
While maintaining similarly low blood damage relative to currently used devices, the P-PAL exhibited high oxygenation efficiency. The maximum blood flow rate for which oxygenation data are available for a wide array of clinically used pediatric devices is 2 L/min. At this blood flow rate, currently used pediatric oxygenators have oxygenation efficiencies of approximately 140–240 ml/min/m2 (as specified by manufacturer literature, see Supplemental Materials, Supplemental Digital Content, https://links.lww.com/ASAIO/A222). The oxygenation efficiency of the P-PAL at the same flow rate is 310 ml/min/m2. This efficiency is comparable to that of the PediPL (approximately 345 ml/min/m2 at 2 L/min blood flow rate).23 A high gas exchange efficiency is important because it allows targeted oxygenation rates to be achieved with a smaller HFM bundle. This in turn enables the device to have a minimized blood-contacting surface area and overall device size.
The high oxygenation efficiency of the P-PAL stems primarily from its HFM bundle geometry. The P-PAL HFM bundle is composed of stacked, circular layers of fibers, resulting in a cylindrical bundle with a small cross-sectional area. During use of the P-PAL, blood flows axially along the cylinder and normal to the axis of the fibers (see Figure 1). Previous modeling and in vitro work has shown that such an HFM bundle design achieves exceptionally high gas transfer efficiencies by maximizing the velocity of blood across the fibers.13 The large relative velocity between the blood and fibers results in a reduction of the diffusional boundary layer at the fiber surface (the primary impediment to the gas transfer process25).
A distinct difference between the P-PAL and PediPL is the manner in which the pump impeller is supported. The PediPL employs the magnetic levitation motor and control system of the CentriMag blood pump,23 whereas the P-PAL utilizes a two-pivot ceramic bearing system. The use of a magnetically levitated impeller theoretically reduces shear stress on the blood,26 but can also add complexity to the device and drive system.27 A recent study comparing the use of two similar blood pumps with and without a magnetically levitated impeller during ECMO found no difference in patient outcomes.28
In conclusion, the P-PAL is a compact, fully integrated blood pump and oxygenator designed to provide adequate respiratory support for children up to 25 kg while enabling patient mobility. In vitro evaluation of the P-PAL has shown pumping and gas transfer capabilities that are more than sufficient to meet the needs of its intended clinical use. The P-PAL also exhibits low device-induced hemolysis and thus is ideal for long-term use. Future work includes in vivo evaluation of the P-PAL during both acute and chronic animal studies. Long-term (> 2 weeks) animal testing will be especially critical for the P-PAL considering the extended duration of its intended use. Incorporation of thromboresistant coatings to the blood-contacting surfaces of the P-PAL is also under development.29
1. Erickson S, Schibler A, Numa A, et al.; Paediatric Study Group; Australian and New Zealand Intensive Care Society: Acute lung injury in pediatric intensive care in Australia and New Zealand: A prospective, multicenter, observational study. Pediatr Crit Care Med 2007.8: 317323.
2. Kneyber MCJ, Brouwers AGA, Caris JA, Chedamni S, Plötz FB. Acute respiratory distress syndrome: Is it underrecognized in the pediatric intensive care unit? Intensive Care Med 200834: 751754.
3. Solomon M, Grasemann H, Keshavjee S. Pediatric lung transplantation. Pediatr Clin North Am 2010.57: 37591, table of contents.
4. Goldfarb SB, Benden C, Edwards LB, et al. The Registry of the International Society for Heart and Lung Transplantation: Eighteenth Official Pediatric Lung and Heart-Lung Transplantation Report–2015; Focus Theme: Early Graft Failure. J Heart Lung Transplant 2015.34: 12551263.
5. Valapour M, Paulson K, Smith JM, et al. OPTN/SRTR 2011 annual data report: lung Am J Transplant 201313: 149177.
6. Elizur A, Sweet SC, Huddleston CB, et al. Pre-transplant mechanical ventilation increases short-term morbidity and mortality in pediatric patients with cystic fibrosis. J Heart Lung Transplant 2007.26: 127131.
7. Puri V, Epstein D, Raithel SC, et al. Extracorporeal membrane oxygenation in pediatric lung transplantation. J Thorac Cardiovasc Surg 2010.140: 427432.
8. Tremblay LN, Slutsky AS. Ventilator-induced lung injury: From the bench to the bedside. Intensive Care Med 2006.32: 2433.
9. Mason DP, Thuita L, Nowicki ER, Murthy SC, Pettersson GB, Blackstone EH. Should lung transplantation be performed for patients on mechanical respiratory support? The US experience. J Thorac Cardiovasc Surg 2010.139: 765773.e1.
10. Maury G, Langer D, Verleden G, et al. Skeletal muscle force and functional exercise tolerance before and after lung transplantation: A cohort study. Am J Transplant 2008.8: 12751281.
11. Turner DA, Cheifetz IM, Rehder KJ, et al. Active rehabilitation and physical therapy during extracorporeal membrane oxygenation while awaiting lung transplantation: A practical approach. Crit Care Med 2011.39: 25932598.
12. Rehder KJ, Turner DA, Hartwig MG, et al. Active rehabilitation during extracorporeal membrane oxygenation as a bridge to lung transplantation. Respir Care 2013.58: 12911298.
13. Madhani SP, Frankowski BJ, Federspiel WJ. Fiber bundle design for an integrated wearable artificial lung. ASAIO J 2017.63: 631636.
14. Madhani SP, Frankowski BJ, Burgreen GW, et al. In vitro
and in vivo
evaluation of a novel integrated wearable artificial lung. J Heart Lung Transplant 2017.36: 806811.
15. Svitek RG, Federspiel WJ. A mathematical model to predict CO2 removal in hollow fiber membrane
oxygenators. Ann Biomed Eng 2008.36: 9921003.
16. Hines AL, Maddox RN. Mass Transfer: Fundamentals and Applications 1985.Englewood Cliffs, NJ: Prentice-Hall.
17. Madhani SP, D’Aloiso BD, Frankowski B, Federspiel WJ. Darcy permeability of hollow fiber membrane
bundles made from Membrana polymethylpentene fibers used in respiratory assist devices. ASAIO J 2016.62: 329331.
18. Svitek RG, Frankowski BJ, Federspiel WJ. Evaluation of a pumping assist lung that uses a rotating fiber bundle. ASAIO J 2005.51: 773780.
19. Narang N, Thibodeau JT, Levine BD, et al. Inaccuracy of estimated resting oxygen uptake in the clinical setting. Circulation 2014.129: 203210.
20. Liu Y, Sanchez PG, Wei X, et al. Effects of cardiopulmonary support with a novel pediatric pump-lung in a 30-day ovine animal model: Evaluation of a pediatric pump-lung. Artif Organs 201539: 989997.
21. Zhang T, Wei X, Bianchi G, et al. A novel wearable pump-lung device: in vitro
and acute in vivo
study. J Heart Lung Transplant 2012.31: 101105.
22. Chang AC, Hanley F, Wernovsky G, Wessel D. Pediatric Cardiac Intensive Care. 1998.Baltimore: Williams & Wilkins.
23. Wu ZJ, Gellman B, Zhang T, Taskin ME, Dasse KA, Griffith BP. Computational fluid dynamics and experimental characterization of the pediatric pump-lung. Cardiovasc Eng Technol 2011.2: 276287.
24. Kameneva MV, Burgreen GW, Kono K, Repko B, Antaki JF, Umezu M. Effects of turbulent stresses upon mechanical hemolysis: Experimental and computational analysis. ASAIO J 2004.50: 418423.
25. Federspiel WJ, Henchir KA. Lung, artificial: Basic principles and current applications. Encycl Biomater Biomed Eng 2004.9: 910921.
26. Moazami N, Fukamachi K, Kobayashi M, et al. Axial and centrifugal continuous-flow rotary pumps: A translation from pump mechanics to clinical practice. J Heart Lung Transplant 2013.32: 111.
27. Kosaka R, Maruyama O, Nishida M, et al. Improvement of hemocompatibility in centrifugal blood pump with hydrodynamic bearings and semi-open impeller: In vitro
evaluation. Artif Organs 2009.33: 798804.
28. Palanzo DA, El-Banayosy A, Stephenson E, Brehm C, Kunselman A, Pae WE. Comparison of hemolysis between CentriMag and RotaFlow rotary blood pumps during extracorporeal membrane oxygenation: CentriMag and ROTAFLOW rotary blood pumps. Artif Organs 2013.37: E162E166.
29. Ye SH, Arazawa DT, Zhu Y, et al. Hollow fiber membrane
modification with functional zwitterionic macromolecules for improved thromboresistance in artificial lungs. Langmuir 2015.31: 24632471.