Search

Thermal modeling of the human eye has been a subject of interest for years; however, the contributions of the existing models to clinical applications were limited. The purpose of this research is to develop the first algorithm that utilizes ocular surface temperature of the human eye (a) to detect the existence of an early stage intraocular tumor, and (b) to provide estimation of the core body temperature (CBT) on the back of the eye. Three versions of thermal models are developed, where two of them are developed for tumor detection and the other for CBT estimation. The first model assumes that the capillary vessels embedded in the vascular region are thermally insignificant, and the CBT is considered to be different from the blood temperature, whereas the second model and the model of the CBT estimation assume the capillary vessels to be thermally significant, causing a temperature difference between the solid structure of the vascular tissue and the blood, and the CBT is considered to be the same as the blood temperature. The thermal analyses of the eye model are performed by solving an inverse heat transfer problem using non-gradient-based optimization methods. The temperature distribution in the eye model is solved using Sundance, a high-performance parallel finite-element solution software, and HOPSPACK, a pattern search algorithm is employed to detect the heat source and to estimate CBT. Numerical results of the tumor detection algorithm show the capability of detecting a hypothetical small choroidal tumor, smaller than the smallest tumor stage classified by the Collaborative Ocular Melanoma Study (COMS). The algorithm is rigorously tested with realistic parameter uncertainties and a random error within the range of ±15 mK of noise-equivalent temperature difference available in commercial devices (e.g., IR cameras), and it has successfully demonstrated its clinical feasibility for patient-specific early stage tumor detection. In addition, the algorithm can qu