DEVELOPMENT OF A PATIENT-SPECIFIC OCULAR MODEL FOR RISK ASSESSMENT OF GLAUCOMA DEVELOPMENT AND PROGRESSION

Abstract

Glaucoma is the leading cause of blindness worldwide. Once the retinal ganglion cell axons are lost they cannot be cured. Therefore, preventative risk assessment measures are important. To be able to perform these tasks, one needs to understand the mechanism behind the axonal blockage that leads to glaucoma. Biomechanical factors are thought to play a role in glaucoma, but the specific mechanism is not explored. In a Finite Element (FE) ocular model, the complex shape of the optic nerve components can be modeled and relevant mechanical quantities, such as stresses and strains due to intraocular (IOP) and/or intracranial (ICP) pressure, can be estimated and their effects assessed. Furthermore, optic nerve head (ONH) morphology and especially lamina cribrosa shape and properties, which are tightly linked to Glaucoma onset and development, vary greatly between individuals. This consequently suggests the development of patient-specific FE ocular models. A method to generate patient-specific ocular models was contrived based on the geometry extracted from Optical Coherence Tomography (OCT) scans. Specifically, retinal layers were segmented using intensity and graph-based algorithms and the segmented layers were then reconstructed with a thin plate spline method. Finally, solid models were created from the reconstructed surfaces and meshed with tetrahedral elements. The geometric details of the generated ONH model correlate well with those of generic models from pertinent literature and special attention was paid to meshing so that the optic nerve region of the ocular model exhibits analysis-suitable element quality. The suggested reconstruction method is semiautomatic and although we aimed to fully capture the complete ONH region, some anatomical structures, which are generally considered relevant and important, could not be extracted from OCT images in vivo. These include the pia arachnoid complex (dura mater and pia mater) that contains cerebrospinal fluid material and is considered to exert ICP. These were handled by carrying out a parametric analysis, using generic models with linear elastic material properties, to establish the degree of importance of the pia arachnoid complex. It was found that pia and dura mater properties can affect post laminar neural tissue and lamina cribrosa biomechanics. As it is currently infeasible to obtain high-quality patient-specific geometries for the pia arachnoid complex in vivo, we embed generic models of the pia and dura mater in our patient-specific ONH model. Viscoelastic material properties of dura mater and sclera were additionally retrieved from physical unidimensional tensile stress-relaxation tests. The influence of viscoelastic material properties at certain levels of ICP/IOP with a generic ocular model was examined, and results indicated, as expected, the importance of viscoelastic properties. Parametric analysis of patient-specific models was performed via the principal component analysis method deriving statistical shape models (SSM). Qualitative, quantitative and biomechanical assessments were performed with the aid of the generated SSM. For the biomechanical assessment, finite element modeling was employed and several patient-specific models, based on SSM shape modes, were generated and tested. We anticipate further enhancements and developments for this approach in the future. Based on the so far obtained results, we find evidence that patient-specific, anatomically detailed 3D ocular models allow for a better understanding of employed biomechanics and can benefit glaucoma risk assessment.