Abstract :
[en] Tissue engineering is an interdisciplinary field that applies the principles of engineering and life sciences towards the development of biological constructs with the ability to heal, improve, or replace damaged tissue. Despite the increasing amount of research which has led to some progress in delivering bench to bedside solutions in the field of bone tissue engineering, the transition rate of these solutions from research to clinics has been slow due to the lack of reproducibility and quality control in the constructed tissues. Computational models, applied in the context of process engineering strategies, are considered as great tools for addressing these issues by providing appropriate means that enable us to predict and optimize the neotissue formation during the in vitro culture. In this thesis, a computational framework is developed investigating the in vitro cell expansion process in 2D tissue flasks and 3D scaffolds in a perfusion bioreactor set-up.
In the first part of this thesis, a data-driven model was built, based on the information of 174 donors with the aim of predicting the population doubling time (PDT) of the donor-derived cells. For this purpose, a Random Forests technique using five cell- and donor-specific input parameters (age, gender, initial seeding, surface area and the previous culture time) was applied. It is showed that the age of the donor has the most important factor influencing the PDT (especially for the first two passages). Furthermore, the prediction error for the PDT is lower than for the current standards used in the laboratory for predicting the PDT.
In the second part of this thesis, a computationally efficient model was developed describing the growth of neotissue (cells and their extracellular matrix) inside 3D scaffolds. The growth of neotissue is modeled as a function of oxygen, glucose, pH level, mean curvature of the neotissue-void interface inside the 3D scaffold, and the shear stress caused by the medium flow that is perfused through the scaffold and the neotissue. Bayesian optimization was used to find the best refreshment strategy that maximizes the neotissue growth during the culture time. Subsequently, a novel cost function considering the cost of labor and culture medium was added to the developed model with the goal of maximizing the neotissue growth and minimizing all associated experimental costs during the culture period. The Multi-objective optimization problem was solved using four evolutionary algorithms and three different settings each. Using the obtained Pareto-front, we were able to choose the most optimal solution to the Multi-objective optimization problem.
Finally in the last part of this thesis, the neotissue model developed in the previous part was further expanded by making a distinction between the growth of cells and production of ECM, as well as considering the effect of a generic growth factor on cell proliferation and differentiation. Using a new cost function taking into account the costs of the growth factor, labor and culture medium, refreshment strategies leading to maximum cell proliferation and minimum overall experimental costs were identified. Therefore, this model could be used to predict cell expansion in 3D scaffolds as an alternative for cell expansion in conventional 2D flasks.
In conclusion, this PhD thesis shows the potential of computational models in investigating the cell expansion and tissue growth in 2D and 3D environments and demonstrates how these models can assist in the achievement of a robust and predictable bioprocess for regenerative medicine applications.
