Retinoblastoma is a rare pediatric cancer affecting the retina, optic nerve, and brain. Accounting for 6% of all cancers in children under the age of five, it is the world's most common primary intraocular childhood malignancy. While survival rates in the U.S. are high, early detection is pivotal to preserving vision: by the time symptoms present, enucleation is often the only option. This proposal seeks to develop IRIS-R, an Intelligent Retinal Imaging Solution to enable earlier detection of Retinoblastoma. This inexpensive, noninvasive screening tool will leverage recent advances in deep learning to detect the tell-tale signs of retinal tumors in near real-time. The models will be integrated with a handheld non-mydriatic fundus imager, providing a reliable, inexpensive, hardware backbone for the screening tool. IRIS-R will be developed with help from retinal specialists, practicing ophthalmologists, and pediatricians to guarantee maximum diagnostic accuracy and clinical usefulness.
