Pascal Verdonck

Ghent University, Gand, Flanders, Belgium

Are you Pascal Verdonck?

Claim your profile

Publications (379)640.34 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Carotid artery stenting (CAS) is an alternative procedure for the treatment of severely stenosed carotid artery lesions in high-risk patients. Appropriate patient selection and stent design are paramount to achieve a low stroke and death rate in these complex high-risk procedures. This study introduces and evaluates a novel virtual, patient-specific, pre-operative environment to quantify scaffolding parameters based on routine imaging techniques. Two patients who underwent CAS with two different sizes of the Acculink stent (Abbott Vascular, Santa Clara, CA, USA) were studied. Pre-operative data were used to build the numerical models for the virtual procedure. Numerical results were validated with post-operative angiography. Using novel virtual geometrical tools, incomplete stent apposition, free cell area and largest fitting sphere in the stent cell were evaluated in situ as quantitative measures of successful stent placement and to assess potential risk factors for CAS complications. A quantitative validation of the numerical outcome with post-operative images noted differences in lumen diameter of 5.31 ± 8.05% and 4.12 ± 9.84%, demonstrating the reliability of the proposed methodology. The quantitative measurements of the scaffolding parameters on the virtually deployed stent geometry highlight the variability of the device behavior in relation to the target lesion. The free cell area depends on the target diameter and oversizing, while the largest fitting spheres and apposition values are influenced by the local concavity and convexity of the vessel. The proposed virtual environment may be an additional tool for endovascular specialists especially in complex anatomical cases where stent design and positioning may have a higher impact on procedural success and outcome.
    The International journal of artificial organs 01/2015; 37(12):928-939. · 1.45 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this study was to reconcile some of our own previous work and the work of others to generate a physiologically realistic numerical simulation environment that allows to virtually assess the performance of BMHVs. The model incorporates: (i) a left ventricular deformable model to generate a physiological inflow to the aortic valve; (ii) a patient-specific aortic geometry (root, arch and descending aorta); (iii) physiological pressure and flow boundary conditions. We particularly studied the influence of downstream geometry, valve size and orientation on leaflet kinematics and functional indices used in clinical routine. Compared to the straight tube geometry, the patient-specific aorta leads to a significant asynchronous movement of the valve, especially during the closing of the valve. The anterior leaflet starts to close first, impacts the casing at the closed position and remains in this position. At the same time, the posterior leaflet impacts the pivoting mechanisms at the fully open position. At the end of systole, this leaflet subsequently accelerates to the closed position, impacting the casing with an angular velocity of approximately -477 rad/s. The valve size greatly influences the transvalvular pressure gradient (TPG), but does not change the overall leaflet kinematics. This is in contrast to changes in valve orientation, where changing valve orientation induces large differences in leaflet kinematics, but the TPG remains approximately the same.
    Annals of Biomedical Engineering 09/2014; · 3.23 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose Over 25% of prevalent U.S. hemodialysis patients receive renal replacement through catheters. Symmetrical tip dialysis catheters have become alternatives to split tip and step tip designs owing to low solute recirculation and ease of positioning. We compared flow characteristics and recirculation of two symmetrical tip dialysis catheters using computational flow dynamics (CFD). Materials and Methods We compared the VectorFlow catheter (with helically contoured distal lumens which offset the vectors of blood entering and leaving the device) with the Palindrome catheter (with symmetrically cross-cut tips within a dual-D configuration). A previously validated CFD model was used to compare these catheters for recirculation (as percent blood flow), Platelet Activation State (PAS) (a computational metric of thrombogenicity) and Helical Flow Index (HFI, to quantify the helical content of the streaming blood) using a steady-state superior vena model assuming catheter flow rates of 400 mL/min. Results Both catheters were associated with a recirculation close to zero. Arterial and venous lumen outflow Platelet Activation State (PAS) (70th percentile) of VectorFlow catheter vs. Palindrome was 1.88 x 10-6 vs. 2.09 x 10-6 (arterial); and 5.08 x 10-6 vs. 1.14 x 10-5 (venous), respectively. Both catheters exhibited a symmetrical venous flow structure away from the catheter tip and coherent helical trajectories entering the arterial lumen, but transitioning rapidly in the VectorFlow catheter to fully developed flow. HFI analysis showed that both the Palindrome and VectorFlow catheters exhibited helical flow, with HFI of 0.032 and 0.024, respectively; the helically contoured lumens of the VectorFlow were also effective in deflecting the venous flow away from the arterial lumen. Conclusion Both VectorFlow and Palindrome catheters showed minimal recirculation rates compared to conventional step tip and split tip alternatives. The VectorFlow catheter was associated with lower Platelet Activation State (PAS) levels and helical deflection of blood from the venous lumen away from the arterial lumen.
    Journal of Vascular and Interventional Radiology 03/2014; 25(3):S21–S22. · 2.15 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In two subjects with a unilateral diaphragmatic paralysis and complaints of dyspnea, a completely different treatment approach was chosen despite similar anatomical and physiological abnormalities. These decisions were supported by the results generated by Functional Respiratory Imaging (FRI). FRI was able to generate functional information with respect to lobar ventilation and local drug deposition. In one subject, it was found that some lobes were poorly ventilated and drug deposition simulation showed that some regions were undertreated. This subject underwent a diaphragm plication to restore the ventilation. In the other subject, it was found that all lobes were still ventilated. A conservative approach with regular follow-up was chosen to wait for spontaneous recovery of the diaphragmatic function. Both subjects improved subjectively and objectively. These cases demonstrate how novel medical imaging techniques such as FRI can be used to personalize respiratory treatment in subjects with unilateral diaphragmatic paralysis.
    Respiratory care. 12/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this work we investigated the possibility to predict the pressure drops across a patient-specific arterio-venous fistula (AVF) by means of an open-source hemodynamics solver aimed at convection-dominated incompressible flows. To account for the very high flow rates that develop in AVFs we considered a wide range of steady input flow conditions (corresponding to Reynolds numbers 100, 200, 550, 1000, 1500, and 2000), and compared with experiments for over 200 flow rates, up to Reynolds 2000. Three meshes for the numerical model, based on a micro-CT acquisition of the in vitro silicon model, were generated, in order to perform a h-refinement study and assess the mesh density allowing to correctly estimate the losses across the anastomosis. For the sake of validation, in addition to pressure assessment, the velocity solutions for Re 550 and 1000 were compared with particle image velocimetry (PIV) acquisitions. Once the solver was validated we also simulated pulsatile input flow conditions to investigate the role of pulsatility in predicting pressure drops. When the finer grid is considered almost all the experimental values for the pressure drop vs. flow measurements are within the standard deviation range of the numerical pressure drops. For the PIV validation, a good agreement is observed between in vitro data and numerical results. The ability to simulate unstable convection-dominated flows in complex 3D geometries is demonstrated and more insight is obtained about the non-common physiological flow conditions induced by fistula creation.
    Cardiovascular Engineering and Technology. 12/2013;
  • Source
  • Source
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose ‐ The purpose of this paper is to use rapid prototyping technology, in this case fused deposition modeling (FDM), to manufacture 2D and 3D particle image velocimetry (PIV) compatible patient-specific airway models. Design/methodology/approach ‐ This research has been performed through a case study where patient-specific airway geometry was used to manufacture a PIV compatible model. The sacrificial kernel of the airways was printed in waterworks™ which is a support material used by Stratasys Maxum FDM devices. Transparent silicone with known refractive index was vacuum casted around the kernel and after curing out, the kernel was removed by washing out in sodium hydroxide. Findings ‐ The resulting PIV model was tested in an experimental PIV setup to check the PIV compatibility. The results showed that the model performs quite well when the refractive index (RI) of the silicone and the fluid are matched. Research limitations/implications ‐ Drawbacks such as the surface roughness, due to the size of the printing layers, and the yellowing of the silicone, due to the wash out of the kernel, need to be overcome. Originality/value ‐ The paper presents the manufacturing process for making complex thick walled patient-specific PIV models starting from a strong workable sacrificial kernel. This removable kernel is obtained by switching the building and the support materials of the FDM machine. In this way, the kernel was printed in support material while the building material was used to support the kernel during printing. The model was tested in a PIV setup and the results show that the airway model is suitable for performing particle image velocimetry.
    Rapid Prototyping Journal 07/2013; 19(5). · 1.16 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Biomechanical research of left ventricular function involves the assessment and understanding of both ventricular wall mechanics and deformation and intraventricular flow patterns, as well as how they interact. Experimental research using hydraulic bench models should therefore aim for an as realistic as possible simulation of both. In previous experimental investigations, wall deformation was studied by means of thin-walled passive experimental models, consisting of a silicone membrane in a closed box, which is squeezed passively by an externally connected piston pump. Although the pump function of these models has already been well established, the membrane deformation remains unpredictable and the effect of muscle contraction – and hence natural wall deformation – cannot be simulated. In this study, we propose a new design of an experimental hydraulic left ventricular model in which left ventricular wall deformation can be controlled. We built this model by a combination of rapid prototyping techniques and tested it to demonstrate its wall deformation and pump function. Our experiments show that circumferential and longitudinal contraction can be attained and that this model can generate fairly normal values of pressure and flow.
    IRBM 06/2013; 34(3):226–234. · 0.38 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this study is to analyze the shape and flow changes of a patient-specific carotid artery after carotid artery stenting (CAS) performed using an open-cell (stent-O) or a closed-cell (stent-C) stent design. First, a stent reconstructed from micro-computed tomography (microCT) is virtually implanted in a left carotid artery reconstructed from CT angiography. Second, an objective analysis of the stent-to-vessel apposition is used to quantify the lumen cross-sectional area and the incomplete stent apposition (ISA). Third, the carotid artery lumen is virtually perfused in order to quantify its resistance to flow and its exposure to atherogenic or thrombogenic hemodynamic conditions. After CAS, the minimum cross-sectional area of the internal carotid artery (ICA) (external carotid artery [ECA]) changes by +54% (-12%) with stent-O and +78% (-17%) with stent-C; the resistance to flow of the ICA (ECA) changes by -21% (+13%) with stent-O and -26% (+18%) with stent-C. Both stent designs suffer from ISA but the malapposed stent area is larger with stent-O than stent-C (29.5 vs. 14.8 mm(2) ). The untreated vessel is not exposed to atherogenic flow conditions whereas an area of 67.6 mm(2) (104.9) occurs with stent-O (stent-C). The area of the stent surface exposed to thrombogenic risk is 5.42 mm(2) (7.7) with stent-O (stent-C). The computer simulations of stenting in a patient's carotid artery reveal a trade-off between cross-sectional size and flow resistance of the ICA (enlarged and circularized) and the ECA (narrowed and ovalized). Such a trade-off, together with malapposition, atherogenic risk, and thrombogenic risk is stent-design dependent.
    Artificial Organs 04/2013; · 1.87 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Abstract Background: As the upper airway is the most important limiting factor for the deposition of inhalation medication in the lower airways, it is interesting to assess how its morphology varies between different postures. The goal of this study is to compare the upper airway morphology and functionality of healthy volunteers in the upright and supine positions during tidal nasal breathing and to search for baseline indicators for these changes. This is done by performing three-dimensional measurements on computed tomography (CT) and cone beam computed tomography (CBCT) scans. Methods: This prospective study was approved by all relevant institutional review boards. All patients gave their signed informed consent. In this study, 20 healthy volunteers (mean age, 62 years; age range, 37-78 years; mean body mass index, 29.26; body mass index range, 21.63-42.17; 16 men, 4 women) underwent a supine low-dose CT scan and an upright CBCT scan of the upper airway. The (local) average (Savg) and minimal (Smin) cross-sectional area, the position of the latter, the concavity, and the airway resistance were examined to determine if they changed from the upright to the supine position. If changes were found, baseline parameters were sought that were indicators for these differences. Results: There were five dropouts due to movement artifacts in the CBCT scans. Savg and Smin were 9.76% and 26.90% larger, respectively, in the CBCT scan than in the CT scan, whereas the resistance decreased by 26.15% in the upright position. The Savg of the region between the hard palate and the bottom of the uvula increased the most (49.85%). In people with a high body mass index, this value changed the least. The airway resistance in men decreased more than in women. Conclusions: This study demonstrated that there are differences in upper airway morphology and functionality between the supine and upright positions and that there are baseline indicators for these differences.
    Journal of Aerosol Medicine and Pulmonary Drug Delivery 03/2013; · 2.89 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVE: We aim to investigate if anatomical and functional properties of the upper airway using computerized 3D models derived from computed tomography (CT) scans better predict obstructive sleep apnea (OSA) severity than standard clinical markers. METHODS: Consecutive children with suspected OSA underwent polysomnography, clinical assessment of upper airway patency, and a CT scan while awake. A three-dimensional (3D) reconstruction of the pharyngeal airway was built from these images, and computational fluid dynamics modeling of low inspiratory flow was performed using open-source software. RESULTS: Thirty-three children were included (23 boys; mean age, was 6.0±3.2y). OSA was diagnosed in 23 patients. Children with OSA had a significantly lower volume of the overlap region between tonsils and the adenoids (median volume, 1408mm compared to 2173mm; p=0.04), a lower mean cross-sectional area at this location (median volume, 69.3mm2 compared to 114.3mm2; p=0.04), and a lower minimal cross-sectional area (median volume, 17.9mm2 compared to 25.9mm2; p=0.05). Various significant correlations were found between several imaging parameters and the severity of OSA, most pronounced for upper airway conductance (r=-0.46) (p<0.01) for correlation between upper airway conductance and the apnea-hypopnea index. No differences or significant correlations were observed with clinical parameters of upper airway patency. Preliminary data after treatment showed that none of the patients with residual OSA had their smallest cross-sectional area located in segment 3, and this frequency was significantly lower than in their peers whose sleep study normalized (64%; p=0.05). CONCLUSION: Functional imaging parameters are highly correlated with OSA severity and are a more powerful correlate than clinical scores of upper airway patency. Preliminary data also showed that we could identify differences in the upper airway of those subjects who did not benefit from a local upper airway treatment.
    Sleep Medicine 03/2013; · 3.10 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: On the one hand the heterogeneity of the circulatory system requires the use of different models in its different compartments, featuring different assumptions on the spatial degrees of freedom. On the other hand, the mutual interactions between its compartments imply that these models should preferably not be considered separately. These requirements have led to the concept of geometrical multiscale modeling, where the main idea is to couple 3D models with reduced 1D and/or 0D models. As such detailed information on the flow field in a specific region of interest can be obtained while accounting for the global circulation. However, the combination of models with different mathematical features gives rise to many difficulties such as the assignment of boundary conditions at the interface between two models and the development of robust coupling algorithms, as the subproblems are usually solved in a partitioned way. This review aims to give an overview of the most important aspects concerning 3D-1D-0D coupled models. In addition, some applications are presented in order to illustrate the potentialities of these coupled models.
    Annals of Biomedical Engineering 12/2012; 41(7). · 3.23 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Purpose: Maturation of an arterio-venous fistula (AVF) frequently fails, with low post-operative fistula flow as a prognostic marker for this event. As pulsed wave Doppler (PWD) is commonly used to assess volume flow, we studied the accuracy of this measurement in the setting of a radio-cephalic AVF. Methods: As in-vivo validation of fistula flow measurements is cumbersome, we performed simulations, integrating computational fluid dynamics with an ultrasound (US) simulator. Flow in the arm was calculated, based on a patient-specific model of the arm vasculature pre and post AVF creation. Raw ultrasound signals were subsequently simulated, from which Doppler spectra were calculated in both a proximal and a distal location. Results: The velocity component in the direction of the PWD-US beam (vPWD), in a centered, small, sample volume, can be captured accurately using PWD spectrum mean-tracking (maximum bias [mB] 8.1%). However, when deriving flow rate from these measurements, a high degree of inaccuracy occurs. First, the angle-correction of vPWD towards the velocity along the axis of the vessel is largely influenced by the radial velocity components in the complex flow field (mB=16.3%). Second, the largest error is introduced when transferring the centerline velocity to the cross-sectional mean velocity without any knowledge of the flow profile (mB=97.7%). Conclusions: In the setting of a forearm AVF, flow estimates based on PWD are hampered by the complex flow patterns. Overall, flow estimation based on centerline measurement, analyzed by mean-tracking of the RF-spectral estimates, under the assumption of a parabolic flow profile, appeared to provide the most reasonable values.
    The journal of vascular access 11/2012; · 1.02 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Although arterio-venous grafts (AVGs) are the second best option as long-term vascular access for hemodialysis, they suffer from complications caused by intimal hyperplasia, mainly located in vessel regions of low and oscillating wall shear stress. However, certain flow patterns in the bulk may reduce these unfavorable hemodynamic conditions. We therefore studied, with computational fluid dynamics (CFD), the impact of a helical AVG design on the occurrence of (un)favorable hemodynamic conditions at the venous anastomosis. Six CFD-models of an AVG in closed-loop configuration were constructed: one conventional straight graft, and five helical designed grafts with a pitch of 105mm down to 35mm. At the venous anastomosis, disturbed shear was assessed by quantifying the area with unfavorable conditions, and by analyzing averaged values in a case-specific patch. The bulk hemodynamics were assessed by analyzing the kinetic helicity in and the pressure drop over the graft. The most helical design scores best, being instrumental to suppress disturbed shear in the venous segment. There is, however, no trivial relationship between the number of helix turns of the graft and disturbed shear in the venous segment, when a realistic closed-loop AVG model is investigated. Bulk flow investigation showed a marked increase of helicity intensity in, and a moderate pressure drop over the AVG by introducing a lower pitch. At the venous anastomosis, unfavorable hemodynamic conditions can be reduced by introducing a helical design. However, due to the complex flow conditions, the optimal helical design for an AVG cannot be derived without studying case by case.
    Journal of biomechanics 11/2012; · 2.66 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Extended nocturnal home hemodialysis has gained renewed interest. However, no removal data for single/double needle (lumen) (SL and DL, respectively) or for low/high blood flow in extended dialysis are available. Therefore, we studied dialysis adequacy in different nocturnal home hemodialysis strategies. Coupling a kinetic with a dialyzer model, we calculated a reduction ratio from pre- to post-dialysis (RR) and total solute removal (TSR) of urea, methylguanidine (MG), β(2)-microglobulin, and phosphate. Simulations were done for dialysis with blood flow Q(b)350 ml/min (DL-4h), extended DL high flow with Q(b)350 (DL-HF-8h) and low flow with Q(b)175 (DL-LF-8h), and SL with Q(b)273 (SL-8h). Compared to DL-4h, TSR was 28-59% larger for DL-HF-8h. TSR was most increased for β(2)-microglobulin (18%) with DL-LF-8h, and for MG (35%) with SL-8h. Furthermore, RRs were equal (DL-LF-8h), higher (SL-8h), and even more increased (DL-HF-8h) for all studied solutes. In the home setting, DL-LF-8h and SL-8h are safe and promising strategies.
    Blood Purification 10/2012; 34(3):219-224. · 2.06 Impact Factor
  • Source
    Journal of Biomechanics 07/2012; 45:S27. · 2.50 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, the influence of the upstream boundary condition in the numerical simulation of an aortic bileaflet mechanical heart valve (BMHV) is studied. Three three-dimensional cases with different upstream boundary conditions are compared. The first case consists of a rigid straight tube with a velocity profile at its inlet. In the second case, the upstream geometry is a contracting left ventricle (LV), positioned symmetrically with respect to the valve. In the last case, the LV is positioned asymmetrical with respect to the valve. The cases are used to simulate the same three-dimensional BMHV. The change in time of the LV volume is calculated such that the flow rate through the valve is identical in each case. The opening dynamics of the BMHV are modelled using fluid–structure interaction. The simulations show that differences occur in the leaflet movement of the three cases. In particular, with the asymmetric LV, one of the leaflets impacts the blocking mechanism at its open position with a 34% higher velocity than when using the velocity profile, and with an 88% higher velocity than in the symmetric LV case. Therefore, when simulating such an impact, the upstream boundary condition needs to be chosen carefully. Copyright © 2012 John Wiley & Sons, Ltd.
    International Journal for Numerical Methods in Biomedical Engineering. 06/2012; 28(6-7).
  • [Show abstract] [Hide abstract]
    ABSTRACT: Fluid mechanical stimuli are known to upregulate cell differentiation and matrix formation. Since wall shear stress plays an important role various studies tried to estimate the scaffold fluid dynamic environment. However, because of the geometrical complexity, nearly all studies created their CFD model based on a submodel of the entire scaffold assuming that the model covers heterogeneity sufficiently. However to the authors' knowledge no study exist providing guidelines in this matter. In a previous study we demonstrated that submodels are influenced by the boundary conditions, inevitable when flow channels are chopped off. For the current study we therefore developed μCT based models of two complete scaffold geometries (one titanium and one hydroxyapatite). Imposing a 0.04 ml/min flow rate resulted in a surface area averaged wall shear stress of 1.41 mPa for titanium and 1.09 mPa for hydroxyapatite. In order to get insight in required model size we subdivided the domain in regions of different size. From our results we propose a model size between 6 and 10 times the average pore size. The wall shears stress should be calculated on a region at least one pore size away from the boundaries. These guidelines could be of use for computationally more costly simulations where it is not possible to simulate the complete scaffold domain.
    Journal of biomechanics 04/2012; 45(9):1586-92. · 2.66 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Hemodialysis patients require a vascular access that is, preferably, surgically created by connecting an artery and vein in the arm, i.e. an arteriovenous fistula (AVF). The site for AVF creation is chosen by the surgeon based on preoperative diagnostics, but AVFs are still compromised by flow-associated complications. Previously, it was shown that a computational 1D-model is able to describe pressure and flow after AVF surgery. However, predicted flows differed from measurements in 4/10 patients. Differences can be attributed to inaccuracies in Doppler measurements and input data, to neglecting physiological mechanisms or to an incomplete physical description of the pulse wave propagation after AVF surgery. The physical description can be checked by validating against an experimental setup consisting of silicone tubes mimicking the aorta and arm vasculature both before and after AVF surgery, which is the aim of the current study. In such an analysis, the output uncertainty resulting from measurement uncertainty in model input should be quantified. The computational model was fed by geometrical and mechanical properties collected from the setup. Pressure and flow waveforms were simulated and compared with experimental waveforms. The precision of the simulations was determined by performing a Monte Carlo study. It was concluded that the computational model was able to simulate mean pressures and flows accurately, whereas simulated waveforms were less attenuated than experimental ones, likely resulting from neglecting viscoelasticity. Furthermore, it was found that in the analysis output uncertainties, resulting from input uncertainties, cannot be neglected and should thus be considered.
    Journal of biomechanics 04/2012; 45(9):1684-91. · 2.66 Impact Factor

Publication Stats

3k Citations
640.34 Total Impact Points


  • 1993–2014
    • Ghent University
      • • Institute of Biomedical Technology
      • • Department of Civil Engineering
      • • Department of Molecular Biotechnology
      • • Department of Structural Engineering
      Gand, Flanders, Belgium
  • 2011
    • Université de Sherbrooke
      • Department of Nuclear Medicine and Radiobiology
      Sherbrooke, Quebec, Canada
  • 2008–2010
    • Universitair Ziekenhuis Ghent
      Gand, Flanders, Belgium
    • Brunel University
      • School of Engineering and Design
      London, ENG, United Kingdom
  • 2009
    • Lake Tahoe Community College
      South Lake Tahoe, California, United States
    • Hogeschool Gent
      • Department of Mechanics
      Gent, VLG, Belgium
  • 2007–2009
    • Stony Brook University
      • Department of Biomedical Engineering
      Stony Brook, NY, United States
  • 2004–2007
    • University of Groningen
      • Department of Biomedical Engineering
      Groningen, Province of Groningen, Netherlands
  • 2003–2005
    • Imperial College London
      • • Centre for Pharmacology and Therapeutics
      • • Department of Chemical Engineering
      Londinium, England, United Kingdom
    • University of Liège
      • Hemodynamics Research Laboratory (HemoLiege)
      Luik, Walloon Region, Belgium