Multi-physics computational models of articular cartilage for estimation of its mechanical and physical properties

Abstract

Recent advances in the realm of computational modeling of complex multiphysics phenomena in articular cartilage enabled efficient and precise determination of articular cartilage properties. However, still accurate quantification of complicated indentation and diffusion processes tying closely with the inhomogeneity of articular cartilage remains challenging. In the present thesis accurate approaches are proposed to capture the mechanical and physical behavior of articular cartilage as faithfully as possible. Finite element models (FE-models) capable of detecting contact between indenter and cartilage surface are developed and applied to spherical indentation process. To predict mechanical and physical properties of cartilage artificial neural networks (ANN) were used and to guarantee the efficacy of the generated ANN they were trained using simulated noisy force-time data. The combination of FE-model and ANN trained with noisy data allowed obtaining cartilage properties robustly. FE-models taking the inhomogeneity of articular cartilage into account were developed and validated and applied to capture neutral (biphasicsolute model) and charged (multiphasic model) solute transfer across articular cartilage in a finite bath experimental setup. Those models could capture the behavior of solute diffusion across cartilage and provide diffusivities and fixed charge densities (FCD) of different cartilage zones. An algorithm consisting of inverse and forward ANNs was developed to obtain the diffusivities of cartilage layers which eliminates the need for computational expertise. The final goal of this algorithm is to introduce a methodology by which properties of cartilage can be determined without any need for computational expertise, which provides a promising opportunity to meet the needs for clinics when it comes to assess the healthiness of articular cartilage during osteoarthritis progression. Effects of bath osmolarity, concentration and charge of solute were investigated using a combination of micro-CT experiments and FE-models. The results suggested that solute charge unlike the osmoalrity and solute concentration had a profound effect on solute diffusion. Porosity and thickness of subchondral plate were identified as two primary factors affecting the diffusion of neutral solutes across subchondral plate. Using a developed multi-zone biphasic-solute model allowed obtaining the diffusivities of cartilage layers as well as subchondral plate. Using a multi-zone biphasic-solute model, we found that overlying bath size, bath stirring and thickness of the formed stagnant layer can substantially influence the diffusion across cartilage. This provides an opportunity to optimally design diffusion experiments.