Fitting ODE models of tear film breakup

TL;DR

This study uses neural networks and mathematical modeling to analyze tear film breakup mechanisms in healthy subjects, providing insights into their causes and potential for classifying dry eye disease.

Contribution

It introduces an automated system combining CNN detection with ODE-based modeling to identify and quantify tear breakup mechanisms in vivo.

Findings

Distribution of TBU causes in healthy subjects aligns with previous data.

Osmolarity increases with evaporation rate and depends on flow.

Potential to classify subjects and establish baselines for dry eye disease.

Abstract

The contribution of different physical effects to tear breakup (TBU) in subjects with no self-reported history of dry eye are quantified. An automated system using a convolutional neural network is deployed on fluorescence (FL) imaging videos to identify multiple likely TBU instances in each trial. Once identified, extracted FL intensity data was fit by mathematical models that included tangential flow along the eye, evaporation, osmosis and FL intensity of emission from the tear film. The mathematical models consisted of systems of ordinary differential equations for the aqueous layer thickness, osmolarity, and the FL concentration. Optimizing the fit of the models to the FL intensity data determined the mechanism(s) driving each instance of TBU and produced an estimate of the osmolarity within TBU. Fits were produced for 467 instances of potential TBU from 15 non-DED subjects. The results showed a distribution of causes of TBU in these healthy subjects, as reflected by estimated flow and evaporation rates, which appear to agree well with previously published data. Final osmolarity depended strongly on the TBU mechanism, generally increasing with evaporation rate but complicated by the dependence on flow. The results suggest that it might be possible to classify individual subjects and provide a baseline for comparison and potential classification of dry eye disease subjects.