Infectious keratitis is corneal inflammation typically caused by bacterial or fungal infection. Infectious keratitis is a major cause of vision impairment and blindness globally, particularly affecting indigent populations. In Southern India, the warm, humid climate, contaminated water, and inadequate healthcare access result in a significantly higher incidence of infectious keratitis compared to other parts of the world. For instance, the incidence of infectious keratitis in the United States is ~11 cases/100,000 people annually and the incidence in Southern India is >100 cases/100,000 people annually.
Keratitis treatment depends on the etiology of the infection; fungal keratitis cases require antifungal medication and bacterial keratitis cases require antibiotics. Microscopic analysis of a corneal culture is necessary for accurate differentiation between bacterial and fungal keratitis as studies have shown that trained cornea specialists can only visually differentiate between bacterial and fungal keratitis with less than 70% accuracy.
Since microscopic analysis is not typically available in rural Indian clinics, we are developing an AI-based mobile application that can make an automated differentiation between fungal and bacterial keratitis in patient eye photographs. We are collaborating with eye hospitals in Southern India to compile a dataset of bacterial and fungal keratitis images to train a convolutional neural network to make the automated differentiation in our application. Such an application will provide patients and caregivers in rural areas with the information required to seek appropriate subsequent treatment for infectious keratitis.