Parameter identification for tear film thinning and breakup
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Studying and modeling the mechanisms that affect the tear film and may cause tear film breakup can lead to better understanding of dry eye disease. Tear film breakup can be observed in vivo using fluorescence imaging; however, many tear film breakup parameter values and ranges cannot currently be measured directly in subject trials. We extract experimental intensity data from fluorescence images of normal, healthy subjects to compare with our model results. Our data fitting scheme is validated with simulated noisy data and shown to be robust. As a crucial step in our approach, we introduce a novel method for estimating localized initial tear film thickness values, which may be useful to optometrists and ophthalmologists. ☐ Evaporation is an important mechanism in tear film breakup that causes relatively slow tear film thinning. We fit a single layer model for tear film breakup with evaporation, osmosis, and flow to simple breakup shapes that form in over eight seconds. The data is taken from normal, healthy individuals. The fit is achieved by minimizing the difference of theoretical and experimental fluorescent intensity over parameters that define the evaporation distribution. Our optimal thinning rates fall near or within experimental ranges and exhibit similar distributions. The fluid flows of our optimal fits captures the inward flow along the tear film that is characteristic of evaporation-driven thinning. The direction of flow along the film is critical and can either oppose or cause tear film thinning. This work provides quantitative understanding of a tear film breakup mechanism that is thought to be closely related to the development of dry eye disease. Further, our results may serve as a baseline that does not currently exist in the literature, and act as a critical building block for clinical understanding of the causes of tear film breakup. ☐ We adapt the time scale of a model with both Marangoni flow and evaporation to fit breakup instances from normal, healthy individuals occurring between one and eight seconds. Marangoni effect-dominated thinning is characterized by outward, strong tangential flow along the film that is induced by an increase in lipid concentration in a “glob” of lipid that thins the film faster than evaporation alone. We compared the results of our fitting procedure for the evaporation-only model described above and this mixed-mechanism model. In general, the thinning rate separates the two models and suggests whether evaporation or the Marangoni effect is the dominant mechanism causing thinning. Our theoretical maximum osmolarity values are largely separated by causal mechanism, and our theoretical minimum tear film thickness values show that the Marangoni effect significantly accelerates thinning of the tear film. These results clearly show that different mechanisms may be important in different breakup instances and that the mechanisms of evaporation and Marangoni flow can cooperate during tear film breakup. Our results give a framework for analyzing and interpreting fluorescence imaging of the tear film that will lead to better data and better understanding of tear breakup and dry eye disease. ☐ Simplified models using ordinary differential equations can capture essential tear film breakup dynamics and be easily fit to a large range of tear film breakup instances. The models are derived by assuming that extensional flow exists within the tear film. We have exact and numerical solutions for various ordinary differential equation models that capture different local fluid flow within the film. We fit the evaporation-only and mixed-mechanism versions to the centers of tear film breakup instances from normal, healthy subjects. In most instances, the simplified models capture the flow direction and suggest which mechanism(s) were active in the breakup instance; the agreement with partial differential equation models is usually good. This work uses a basic science approach to capture mechanisms of tear film breakup in the simplest way possible and validates the procedure against more complicated models so that it may be automated. ☐ Our research team is actively engaged in work to fully automate a system to identify and fit models for tear film dynamics to experimental image data taken from healthy, normal subjects. In this dissertation, we present our current work that focuses on the fitting part of the automatic system. We show preliminary results of fitting the simplified models to automatically identified breakup instances that were selected using a convolutional neural network trained on over 40,000 images. We fit more automatically-identified experimental tear film breakup instances with our automatic fitting procedure than the rest of the work in this dissertation combined. Our work introduces a shift to a data-oriented modeling approach for tear film dynamics, and may be important for generating ranges of clinical parameters that could be useful to vision scientists and eye doctors. The ability to sample much larger numbers of tear film breakup through faster data processing and analysis from these methods will be crucial to a better understanding of the dynamics of tear film breakup in vivo at a population level, and should lead to better informed treatment of dry eye disease.
Description
Keywords
Dry eye, Fluorescence imaging, Optimization, Tear breakup, Tear film