Preventing corneal blindness caused by keratitis using artificial intelligence.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Medical Imaging

Background:

• Keratitis is a primary cause of corneal blindness globally.
• Early detection and treatment are crucial for preventing vision loss.
• A shortage of ophthalmologists hinders timely keratitis diagnosis, particularly in underserved regions.

Purpose of the Study:

• To develop and evaluate a deep learning system for automated keratitis classification.
• To assess the system's performance across various imaging devices, including smartphones.
• To provide a tool for early keratitis detection to mitigate blindness.

Main Methods:

• A deep learning model was trained on 6,567 slit-lamp images.
• The system classified images into keratitis, other corneal abnormalities, and normal cornea categories.
• Performance was evaluated using Area Under the Curve (AUC) metrics.

Main Results:

• The system achieved high performance (AUCs > 0.96) on diverse digital slit-lamp and smartphone images.
• Sensitivity and specificity in keratitis detection were comparable to experienced ophthalmologists.
• The AI demonstrated robust performance across different camera types.

Conclusions:

• The developed deep learning system shows significant potential for accurate keratitis diagnosis.
• It can be deployed on both specialized cameras and smartphones, enhancing accessibility.
• This technology can facilitate early intervention, reducing the incidence of corneal blindness.