Sai 

As artificial intelligence becomes an increasingly promising tool for image analysis, there is a potential to utilize this technology in order to expedite early detection of diseases. Approximately 60 million people worldwide suffer from Glaucoma, a disease in the eye, 50% of these cases go unnoticed until later stages, when vision loss occurs. Optical Coherence Tomography (OCT) reports remain the main imaging technique used to diagnose this disease. However, the reports can only be effectively analyzed by experienced clinicians. There is great potential in developing a deep learning model to highlight important regions that novice clinicians can focus on. In order to achieve this, convolutional neural networks (CNNs) were employed. First, these networks convert image data into numbers. The network then looks for patterns in the converted data. They trained two CNN models based on eye tracking data collected from expert clinicians when diagnosing glaucoma. Of the two prebuilt models, the U-Net model outperformed ResNet in precision, as it preserved spatial details, which made it more adaptive, while making fewer but more accurate predictions. ResNet identified most of the salient regions; however, U-Net was more precise, predicting fewer regions with greater accuracy. This makes U-Net more clinically useful, as highlighting entire regions indiscriminately would provide little diagnostic value. These findings suggest that AI-assisted detection could enable earlier disease identification, potentially increasing patients’ health and increasing the access to Glaucoma care.