Skip to main navigation menu Skip to main content Skip to site footer

Towards Precision in Ocular Surface Temperature Measurement: Improvements in the ThermOcular System's Image Segmentation

Abstract

Ocular Surface Temperature (OST) assessment is an important biomarker for ocular health evaluation. Infrared (IR) thermography, prized for its precision and non-invasiveness, stands out in this domain. The ThermOcular system, is a tool for assessing OST and combines machine vision and thermal imaging for corneal segmentation to locate the region of interest. We expanded the ThermOcular's capabilities using state-of-the-art semantic segmentation models to improve and extend the identification of additional ocular components, specifically the pupil, iris, and sclera. Drawing from two datasets, our exclusive 'Apricot' and the renowned TEyeD, and utilizing OCRNet in conjunction with transfer learning, significant enhancements in segmentation performance were achieved . As a result, the upgraded ThermOcular system presents unparalleled precision in OST assessments, paving the way for innovative advancements in ocular health diagnostics.
PDF