We present a stochastic approach to model the mechanical deterioration of the reinforcing microstructure of the human cornea. The fundamental structural micro-components of the stroma, collagen and crosslinks, are assembled deterministically into an elementary trusswork cell, multiply repeated and distorted to form a three-dimensional shell with the shape of a cornea. The spatial orientation of the collagen-like elements of each cell is thus characterized stochastically with a non correlated random angle field, obeying an assigned probability density function, leading to a non-deterministic structural stiffness. It follows that the mechanical response of the model to the action of deterministic forces equivalent to the intraocular pressure is stochastic due to the random spatial orientation of collagen fibers. The deterioration of the mechanical stiffness of the collagen components is described through a scalar variable field, evolving in space and in time, representative of a progressive damage which causes heterogeneity and asymmetric behavior. The damage variable acts in two ways on the global stiffness: (i) by reducing the stiffness of the collagen components; (ii) by modifying the dispersion coefficient of the probability density function. The equilibrium equations of the damaging model are solved at discrete time steps, with a fully explicit solution scheme, by means of the stochastic finite element improved perturbation method. The results show that when the collagen fibril stiffness reduces to 10% of the healthy value, as expected in the case of the vision-impairing condition known as keratoconus, the displacement field due to intraocular pressure is significantly affected in terms of both average and variance distributions. This effect confers a typical conical shape to the cornea. In particular, the analysis shows that high values of the response variances are confined in the keratoconus area, which agrees with a high level of uncertainties due to loss of fibril organization and thickness reduction under pathologic conditions.
Read full abstract