Retinal disease is a leading cause of blindness worldwide. Indirect ophthalmology is the current standard of treatment for diseases like diabetic retinopathy, retinopathy of prematurity, and retinal detachment, but must be performed by an ophthalmologist. A telemedical solution for diagnosis is required in regions that do not have an on-site ophthalmologist.
What it does
Our project provides a telemedical solution to the remote diagnosis of retinal disease in the absence of an on-site ophthalmologist. Using a low-cost camera, we programmed a complete telemedicine workflow that included camera warping calibration, VLC streaming, image capture to cloud, Dropbox integration, and custom retinal montaging for remote, holistic diagnosis. In addition, we explored the use of Deep Convolutional Neural Networks in classifying retinal images using Tensorflow and the Kaggle Diabetic Retinopathy Dataset.