VisionAI - Human Eye Fundus Screening System Based on AI Deep Learning Technology

Abstract

The recent development in artificial intelligence contributed to the utilization of information and communication technologies in medical fields. In ophthalmology, the fundus photograph decoding technology is receiving wider attentions as it can easily detect the retinal disorders without having to embrace side effects or inconveniences that follow with pupil dilation test. As AI-based diagnostic technologies can effectively discover disorders in optic nerves, optic layers in retina, retinal vessels, it can be useful in early detection and health checks. This study therefore develops a model which classifies and analyzes 24,000 fundus photographs into four categories (normal, cataract, glaucoma, diabetic retinopathy) based on diagnostic data. The model is further realized into a website which will contribute to effective diagnoses of fundus diseases. Convolutional neural network (CNN, specialized in image processing) is applied as a learning model and EfficientNet was used to configure the network. Hyperparameter optimization was used for tuning, and the developed model is later realized as a public webpage. For the enhancement of performance, this model would necessitate extensive datasets and more intricate classifications of fundus diseases through the collaborative research with medical institutions. The author anticipates more prompt diagnosis and treatment for patients with reduced accessibility and quicker diagnosis for medical professionals.