A Review of Deep Learning in Medical Image Recognition for Glaucoma

Abstract

Glaucoma is a leading cause of blindness globally, making early diagnosis and treatment critical for vision preservation. The advancement of deep learning has accelerated progress in medical image recognition, which offers new possibilities for intelligent diagnosis. This paper outlines the research background and significance, detailing techniques such as Convolutional Neural Networks (CNNs), image segmentation, and feature extraction. It reviews achievements in optic disc and cup segmentation, glaucoma classification, and model optimization, specifically discussing techniques such as transfer learning, data augmentation, and attention mechanisms to enhance model robustness, while noting limitations regarding data standardization, model generalization, and interpretability. Currently, computer-aided diagnosis demonstrates high accuracy and clinical potential. Future efforts should focus on refining datasets, developing lightweight models, and advancing interpretability and multi-modal learning to support glaucoma prevention and treatment.