Keratoconus Detection Using Deep Learning

The eye is considered as one of the most complicated organs of the human body, with various ailments that can impact it. Eye diseases are any condition that disrupts its function, leading to vision problems or blindness caused by genetics and aging to injuries, infections, and environmental factors. Keratoconus is a condition where the cornea becomes thinner and bulges out, causing vision and visual impairment, and must be detected early to prevent potential complications. This paper used real data taken from TOMEY TMS-5 to build a dataset for convolutional neural networks (CNNs) GoogleNet algorithms to classify either a Keratoconus or a normal case. The suggested models provided correct decision up to 85% accuracy which is a good result compared to other approaches using raw data to avoid limiting decision to some parameters. The built dataset contains a group of patients (females and males) whose ages ranged from 14-75 years.