Zhong, Lan2018-12-042018-12-042018http://udspace.udel.edu/handle/19716/23959Rapid tear film thinning has been frequently observed in vivo. The tear film can thin dramatically within one second. However, the dominant mechanisms are unknown. There are different arguments why rapid tear film thinning occurs. In this dissertation, we hypothesized that a non-uniform lipid layer drives tear film thinning; thicker lipid (glob) spreads to its surrounding thinner region, which induces a strong tangential flow and thins the aqueous layer. To test this hypothesis, we built lipid-driven thin film models using lubrication theory. The models successfully captured a very strong tangential flow as well as the short time scale. One successful prediction is that tear film breakup (TBU) occurs beneath a small glob and around the edge of a large glob; these match well with in vivo experimental results. ☐ We then adapted our lipid-driven model by adding equations for solute (salt ions and fluorescein) transport to investigate osmolarity and fluorescein concentration distribution. In rapid thinning, osmolarity and fluorescein concentration remains approximately constant. With the computed fluorescein concentration and aqueous layer thickness, we estimated fluorescein intensity, which is often used to visualize the tear thinning in clinical settings. Fluorescein intensity is proportional to aqueous layer thickness if tear film breakup time (TBUT) is less than 4 seconds. For tear film breakup (TBU) longer than 4 seconds, an initial fluorescein concentration less than about 0.2% can capture rapid tear thinning accurately. We made a close comparison between our simulation results (predicted fluorescein intensity) with in vivo experimental results, and the simulation matches the time scale and physical dimensions very well. ☐ Evaporation has been believed to be a major mechanism of TBU that develops over long times; our mathematical models in this thesis showed that the lipid layer can play an essential role in rapid TBU. This additional understanding of TBU provides insights about better imaging methods and treatment of lipid-driven TBU. ☐ To date, processing experimental images has been a very labor intense task. Thousands of high resolution microscopic images of the lipid layers of several thousand subjects were taken by a research group from The Ohio State University to study lipid layer structure. However, most of these images share limited and similar patterns. To better analyze these images, we built a classification model under the framework of Bag of Features using a small portion of the whole dataset. The classification model is then utilized to classify the remaining images. The model achieved an accuracy of 0.82 on the test dataset and highly reduced the workload of analyzing lipid layer images.Pure sciencesApplied sciencesTear film breakup (TBU)Thesis1077286308https://doi.org/10.58088/hkqm-fs312018-10-17en