Development of computer methods in the investigations of ocular surface temperature by infrared thermography

Ocular surface temperature (OST), which nowadays is primarily captured using infrared (IR) thermography on human subject, refers to the temperature measured at the ocular surface enclosed by palpebral fissure. Hitherto, numerous ocular physiologies and pathologies were unveiled by the investigations in ocular surface temperature despite the absence of sophisticated and consistent analytic methods in the field. The analyzing measures developed in literature on ocular surface temperature are monotonous and inconsistent. First-order spatial statistical analysis was widely adopted in investigations, and ocular surface was fragmentally studied. Corneal center was subjectively located, and with that temperatures of a number of points or regions were picked up to perform statistical analysis. The inconsistency in anatomical localization and the monotony in analysis have admittedly constrained the capability ocular thermography could have achieved. In this report, second order spatial statistical analysis—one of the textural analysis—was introduced to extract thermal texture and looked into the aging effect. Ocular surface was entirely investigated using a warping technique to reveal the role of aging physiologies on the surface temperature, and estimate the tear evaporation rate based on the recorded ocular thermographic sequence. Eyelashes were removed to improve data quality, and automated ocular, corneal localization were developed in the attempt to substitute manual localization, or introduce consistency in the determination of corneal region. Elder subjects were determined to exhibit dissimilar thermal texture with younger subjects, and their ocular surfaces were cooler as a consequence of weaker corneal sensitivity; also, the corneal tear evaporation of the elderly was observed to be higher. Tear film was found to play a major role in the temporal topographical variation, and the absence of this thin layer might result in warmer ocular surface. Automated localizing algorithms for both thermal image and thermographic sequence achieved considerable accuracy in the delineation of ocular and corneal area, and were feasible alternatives to manual localization.DOCTOR OF PHILOSOPHY (MAE