Glaucoma is the leading cause of irreversible blindness globally, with its diagnosis remaining a complex challenge due to the variability and intricacy of clinical assessments. Recent progress in deep learning-based artificial intelligence (AI) offers promising avenues for automating glaucoma detection. This review provides a comprehensive survey of state-of-the-art deep learning methods applied to glaucoma diagnosis using data from fundus photography, optical coherence tomography (OCT), and visual field tests. Relevant studies published between 2020 and 2026 were selected from databases such as ScienceDirect, Google Scholar, and IEEE Xplore, based on well-defined inclusion criteria and search keywords. The review categorizes key deep learning approaches into several groups: convolutional neural networks (CNNs), autoencoder-based models, attention mechanisms, generative adversarial networks (GANs), geometric deep learning frameworks, Transformer-based architectures, reinforcement learning models, self-supervised learning techniques, recurrent neural networks (RNNs) including LSTMs, diffusion models, and hybrid configurations. Notable challenges include the scarcity and lack of diversity in available datasets, difficulties in integrating multimodal data, and the limited interpretability of AI models from a clinical perspective. This article aims to support AI researchers in selecting appropriate deep learning frameworks for glaucoma detection by considering factors such as data characteristics, model architecture, and clinical applicability.
Yadav et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: