Recent projects

Integration of orthotropic material parameters into finite element modeling of left ventricular mechanics

Student: Koen Franse

Supervisors: Peter Bovendeerd, Ir. Emanuele Rondanina

Nowadays most finite element models of the left ventricle (LV) are based on transversely isotropic material parameters, while experimental data point out difference in stiffness in the sheet and sheet-normal direction, indicating orthotropic parameters. To compare the effect of these two material characteristics on the time course of the strain in the LV wall, a model by Bovendeerd et al. was adapted. Orientation of the sheets was obtained from experimental data by Usyk et al.; orthotropic material parameters were taken from a FE model study by Schmid. The integration of the orthotropic parameters into the model did not show any improvement yet towards experimental strain results presented by Ashikaga et al. Since the hemodynamics of the adapted model deviated much from the original model results, the used orthotropic parameters were not directly suitable for this specific model and need to be optimized before proper comparisons can be performed.

Preload-Afterload experiment in FEniCS Project finite element package

Student: Lex van Houts

Supervisors: Peter Bovendeerd, E. Chen 

To gain basic and better understanding of (patho-)physiological events of the heart, mathematical models of the left ventricle were developed. The use of these models could give basic and better understanding of (patho-)physiological events of the heart. These models can be used to translate clinical measurements into information, which cannot be measured but is of great diagnostic value. Recently researchers of the cardiovascular biomechanics research group of Eindhoven University of Technology decided to switch from the current finite element (FE) package SEPRAN to another FE package, FEniCS project. The aim of the present study is to replace the contraction model used in FEniCS project implementation by the contraction model used in the SEPRAN implementation to enable comparison of the FEniCS project and SEPRAN results. This was done by creating three models, a 1D-, 3D-, and an FE model. The 1D model was for didactic purpose only, while the conclusion is based on the results of the 3D and FE model. With these models the preload-afterload experiment of Brutseart and Sonnenblick was simulated. The obtained graphs and parameter values were compared. The results show that the overall trend of the graphs of the 3D and FE model are almost identically. The values of the parameters differ in the range of 0.001 μm to 0.005 μm in length of the sarcomere and contractile element. The differences in the stresses are in the range of 0.01 kPa to 0.08 kPa. From the results can be concluded that the contraction model in the FEniCS implementation is replaced by the contraction model of the SEPRAN implementation, which enables comparison of FEniCS and SEPRAN results. 

The influence of the respiratory mechanics on the cardiovascular system

Student: Tessa M. van Haaften

Supervisors: Peter Bovendeerd, Ir. Tilai Rosalina

Several cardiovascular models have been proposed by TU/e, however respiratory models were too simple to describe the oxygen uptake. In this study a model has been developed to investigate the influence of the mechanical aspects of the respiratory model on the hemodynamics. The model is able to capture the affecting pleural pressure on the pulmonary circulation, which decreases the oxygen concentration during inspiration and increases during expiration.

A mathematical model of the baroreflex

Student: Marissa van Elzen

Supervisors: Peter Bovendeerd, Ir. Tilai Rosalina

In-vitro determination of heart valve regurgitation using ink based video densitometry

Student: Koen Janssens

Supervisors: Marcel Rutten

Valve Regurgitation can be a life threatening condition affecting almost 10 million people in the US alone. In severe cases surgery is required in order to restore the cardiovascular system to physiological values. Good quantification of the regurgitation fraction can aid in deciding whether to perform surgery or not. Current methods are imprecise in quantifying the severity of regurgitation and video densitometric angiography can be a novel method in order to accurately determine a regurgitation fraction. The accuracy of video densitometry will be tested using a mock-up simulation of the left heart chamber under physiological conditions. Instead of a CT-scanner, a high speed camera and ink was used to register fluid ow across a synthetic valve of which the regurgitation fraction could be varied. Ink was injected using a Acist CVi injection system and pigtail catheter resembling a clinical setting. The pigtail was placed downstream of the aortic valve, injecting ink into the set-up's equivalent of the aortic root. The principle is based on darkening of the aortic root through ink injection. Healthy valves would prevent fluid from leaking into the set-up's equivalent of the ventricle, leaving fluid there transparent. Whereas regurgitating valves would not, darkening fluid in the ventricle. Fluid darkening in the ventricle should be directly proportional to the regurgitant volume of water. Comparing fluid darkening of both the aortic and ventricular part of the circulatory system allows for computation of the regurgitation fraction. Results showed good accuracy and consistency for moderate regurgitation fractions and an overestimation of around 10% for mild and severe fractions. The point in time where injection was triggered did not seem to have large effects on the results nor did sampling frequency. Conclusively this method seems to work reasonably in a clinical setting for the quantification of moderate regurgitation fractions, although some effort can still be made in order to improve accuracy.

Review of preclinical models used for the evaluation of anticoagulants to prevent prosthetic heart valve thrombosis

Student: Rick Stolk

Supervisors: Marcel Rutten

After prosthetic heart valve replacement patients carry the risk for thromboembolic events. Despite antithrombotic treatment they are at risk of developing bleeding complications due to the limitations of this antithrombotic treatment. The current anticoagulant treatment of prosthetic heart valve thrombosis (PHVT) is associated with an incidence of up to 5.7% per patient year, being one of the major causes of mortality and morbidity. It is estimated that approximately 850,000 patients would require a prosthetic valve replacement in 2050, which underscores the need for safe and effective anticoagulant therapies. Understanding of the pathophysiology of PHVT is essential for the development of novel anticoagulation therapies. Consequently, the in vitro and in vivo models of PHVT play an essential role in the development and evaluation of novel anticoagulation therapies. However, every model has advantages and limitations. Predominantly, two in vitro models have been used for the assessment of anticoagulation on PHVT, where the thrombogenic characteristics of the used materials limit a proper and clean analysis. A porcine model is preferred for in vivo evaluation of anticoagulants due to their resemblance of the human coagulation system. Several models, in vitro and in vivo, will be discussed which have been used to study the effectiveness of anticoagulants for thromboprophylaxis of prosthetic heart valves.

Determining aortic regurgitation with thermodilution- Proposing a new method to quantify perivalvular leakage

Student: Ellen de Boer

Supervisor: Marcel Rutten 

Perivalvular leakage (PVL) often occurs after transcatheter aortic valve replacement (TAVR). PVL is quantified in terms of aortic regurgitation (AR). There are many techniques to determine PVL, however, cardiologists lack a simple, quantifying method to determine the grade of PVL severity. This study aims to investigate a new method to quantify AR based on temperature. Therefore, an in-vitro model is used that mimics the left ventricle of the heart and the systemic circulation. In this model, the pressure and flow in the aorta are measured. Next to this, the temperature is measured at two points (5.5 and 3.5 cm) upstream from the aortic valve and at one point (3.5 cm) downstream from the valve. Water at 37˚C is used to mimic the blood. With a computer-triggered injection pump a bolus of cold saline is injected. After injection, the ratio of the temperature difference over the length of the valve is calculated to determine the regurgitation fraction (RF). This temperature based RF is compared with the flow-based RF, which serves as a reference value. Based on this comparison, the temperature based RF after a bolus injection does not provide a correct regurgitation fraction. For further research, it is suggested to focus on continuous injection of cold saline and to compare the temperature based RF with video densitometry methods.

Determination of aortic regurgitation through thermodilution- Proposing a new method to quantify aortic regurgitation

Student: Pim Joostens

Supervisors: Marcel Rutten 

After transcatheter aortic valve replacement paravalvular leakage can occur. Cardiologists have no simple quantifying method to determine the level of paravalvular leakage severity, which would give an indication for further medical procedures. This study aims to investigate the viability of determining the severity of aortic regurgitation through temperature measurements after injecting a cold solution downstream of the aortic valve. Temperatures up- and downstream from the aortic valve would be monitored throughout injection and the constant temperature after injection would be subtracted from the temperature at the start of the injection to calculate a temperature difference. The ratio between the temperatures differences before and after the aortic valve would then said to be the regurgitation fraction. A model of the functional hearth circulation was used to compare a newly proposed temperature method to an already validated flow measurement. Matching the results for the RF would indicate whether aortic regurgitation is quantifiable through the proposed method. From this comparison it became clear that a relation between both methods exists, however aortic regurgitation was not quantifiable though the newly proposed method. Further research could focus on in-vivo experiments and positioning of the thermocouples. 

Determination of cardiovascular changes as a result of pulsating LVAD phase shifting

Student: Teun van de Meerakker

Supervisors: Marcel Rutten

-no abstract-

Analysis of the tidal wave in blood pressure waveform with a 1D wave propagation model

Student: Nikki Valckx

Supervisors: Frans van de Vosse, Mariëlle Bosboom 

Understanding waveforms and wave reflections is critical when you want to use blood pressure waves to obtain insight into the (mal)functioning of the arterial system. The radial pressure waveform can be determined non-invasively. In the radial pressure waveform two or three peaks occur. The number of peaks depends on whether the tidal wave is visible or not. In earlier analysis a dataset showed a clear tidal peak, especially in female subjects. In this research a 1D model is used to simulate the radial pressure waveform. Parameters as arterial stiffness, heart rate and vessel wall thickness are adjusted to study the effect on the tidal wave in the radial pressure waveform. The 1D model captures previously reported waveform behavior quite well. However, generating a clear tidal wave as was found in the measured dataset could not be achieved. The dataset shows remarkable results when separating the results on date of measurement. Due to these results and the missing information about the measurement setting, it is suggested to do new measurements. For future research it would be better to have subjects in a larger age range and to measure the radial pressure waveform of one subject on multiple days.

The effect of myocardial bridging on the hemodynamics in a stenotic coronary artery

Student: Bing Lin

Supervisors: Mariëlle Bosboom, Ir. Kujtim Gashi

Myocardial bridging (MB) is an anomaly in the coronary arteries which is ignored in the clinic due to its non-malicious presence. However MB in combination with stenosis can be dangerous. In this report a MB in a healthy and a stenotic left anterior descending (LAD) coronary artery is added to an existing fluid-structure interaction (FSI) model for a straight tube. The Arbitrary Lagrange-Euler (ALE) method is used to solve this FSI problem. The findings of this report suggest that the length of a MB is a more important variable to consider for a healthy LAD artery with MB than its location or its depth. A bridged segment should be extra carefully be screened for stenosis because the co-existence of stenosis and MB in the same segment results in a more narrowed LAD coronary artery. For the further development of this model, more correct parameters are needed. Total resistance of the three-element Windkessel model representing the pressure in the myocardial tissue and the length of the cardiac cycle has been proposed to be the two relevant parameters to consider in the future research.

Comparing myocardial resistance estimation in virtual fractional flow reserve modeling to clinical measurements

 Student: Catherine Taelman

Supervisors: Mariëlle Bosboom, Marcel van ‘t Veer, Ir. Kujtim Gashi

A stenosis is the narrowing of a coronary artery, which results in ischemia. To assess the severity of a stenosis, fractional flow reserve (FFR) is measured. However, this is an invasive measurement that involves risks and besides that, procedural costs are high. Virtual fractional flow reserve (vFFR) is a promising alternative that calculates the FFR from non-invasive medical data by using a FEM CDF model. However, boundary conditions need to be determined so that they reflect the in vivo physiology of the coronary arteries best. One important boundary condition is the myocardial resistance.

In previous work, myocardial resistance (Rmyo) has been estimated from an equation that matches the virtual FFR to the measured FFR. This method of estimating the myocardial resistance was now compared to the myocardial resistance obtained from clinical measurements. Coronary angiography data of 8 patients were included for this comparison. The resulting measured Rmyo ranged from 3.00 106 to 6.1 106 kPa s/m3 with a mean of 4.00 106 kPa s/m3, whereas the virtual (estimated) Rmyo ranged from 0.8 106 to 2 106 kPa s/m3 with a mean of 1.56 106 kPa s/m3. An explanation for the lower virtual Rmyo values is the use of generic aortic pressure and viscosity values in the model. Therefore, the influence of blood viscosity on the myocardial resistance was examined as well and it was found that a higher blood viscosity resulted in a higher virtual Rmyo.

It can thus be concluded that the method used to estimate the myocardial resistance is good, but still needs to be improved by making boundary conditions like blood viscosity and aortic pressure patient-specific. In this way, the myocardial resistance estimate will become more realistic.

Relationship between myocardial resistance and coronary artery length and diameter

Student: Ariënne Baijense

Supervisors: Mariëlle Bosboom, Marcel van ‘t Veer, Ir. Kujtim Gashi 

Cardiovascular diseases, and within them coronary artery diseases, are worldwide a very big problem. In coronary artery disease a vessel is narrowed due to stenosis. The severity of the stenosis can be assessed from angiography and by calculate FFR, the ratio between distal and proximal blood pressure. Not only is the coronary artery affected in case of a stenosis, but also the tissue distal to the stenosis. The resistance is used to determine to what extent the myocardial bed is affected. This can be done by calculating the resistance and compare it to normal values. In case the myocardial resistance depends on diameter, length or mass these parameters can be used to determine the severity of the stenosis in future. This relationship is looked for with the use of twenty datasets from which the resistances are calculated.

Myocardial resistances are between 2.47·106 and 9.34·106 kPa·s·m-3. Furthermore for eight patients angiograms are segmented to find branch lengths. Lengths were found between 99 mm and 180 mm. With the datasets a positive relation between myocardial branch length and resistance was found. However, it is still not possible to state a specific relation between myocardial mass and resistance, because of inaccuracies of the determination and the impossibility to segment accurate and precise from the sets of angiograms.

Comparing the visualisation of blood with ultrasound imaging to photoacoustic imaging

Student: Jessica Burger

Supervior: Richard Lopata  

Mesh-Based Monte Carlo Simulation of the Photon Distribution in Photoacoustic Imaging

Student: Osman Akdag  

Supervior: Richard Lopata

Photoacoustic imaging of plaques PA Simulation of Plaques Using the K-Wave Toolbox

Student: Femke Vaassen 

Supervior: Richard Lopata

Creating a tool chain for simulation of photoacoustic imaging

Student: Max van Wanrooij 

Supervior: Richard Lopata 

Simulating the heart with FEM and Field II

Student: Amber Meeder

Supervior: Richard Lopata

Tracking improvement of the supported heart

Student: Dieter Timmers

Supervior: Richard Lopata

Carotid 3-D reconstruction using 2-D ultrasound an a probe tracker

Student: Janneke Cruts

Supervior: Richard Lopata 

Modeling of abdominal aortic aneurysms using 4D ultrasound

Student: Arjet Nievergeld

Supervior: Richard Lopata

Mechanical characterization of healthy carotid arteries using 2D ultrasound

Student: Elcke Vloedgraven

Supervior: Richard Lopata

Characterization of aortas (experimental) for validation of AAA elastography

Student: Jesper Schouren

Supervior: Richard Lopata

Validation of photoacoustic vasculature imaging with ultrasound

Student: Fleur Jacobs

Supervior: Richard Lopata

Optimization of signal filtering in photoacoustic vasculature imaging

Student: Luuk Dappers

Supervior: Richard Lopata

Photoacoustic and elastography measurements on the skin, a phantom study

Student: Sascha Massop

Supervior: Richard Lopata