Frontier in Medical & Health Research
DEEP LEARNING AND ARTIFICIAL INTELLIGENCE IN GLAUCOMA DIAGNOSIS: A SYSTEMATIC REVIEW
pdf

Keywords

Glaucoma; Artificial Intelligence; Deep Learning; Explainable AI; Biomedical Engineering; Optical Coherence Tomography; Systematic Review.

How to Cite

DEEP LEARNING AND ARTIFICIAL INTELLIGENCE IN GLAUCOMA DIAGNOSIS: A SYSTEMATIC REVIEW. (2026). Frontier in Medical and Health Research, 4(6), 23-35. https://fmhr.net/index.php/fmhr/article/view/3024

Abstract

Background

Glaucoma is one of the most notable causes of irreversible blindness globally, whereby early detection is usually hampered due to the complexity of disease presentation and inconsistency with the traditional methods of diagnosis. The latest development in Artificial Intelligence, Deep Learning, and Biomedical engineering has brought new perspectives and possibilities of automated detection and progression prediction, as well as intelligent decision-support in glaucoma care.

Objective

To synthesize existing evidence on the role of artificial intelligence, deep learning, and engineering innovations in glaucoma diagnosis and management, including the diagnostic performance, explainable artificial intelligence, translational applications, and future clinical implementation.

Methods

The systematic review was performed based on PRISMA guidelines. The search was performed on the electronic databases of PubMed/MEDLINE, Scopus, Web of Science, Embase, Cochrane Library, and Google Scholar of the published studies dated between 2020 and 2026. The included studies had to be eligible by applying artificial intelligence or deep learning in glaucoma diagnosis or treatment using fundus imaging, optical coherence tomography, visual field analysis, or multimodal techniques. Synthesis of data was done in a narrative way and methodological quality was evaluated with the help of QUADAS-2.

Results

A total of 21 studies were included, including systematic reviews, meta-analyses and diagnostic studies that assess the use of artificial intelligence in glaucoma care. Deep learning models showed good diagnostic performance in imaging modalities with some studies reporting high sensitivity, specificity and comparable accuracy to experts. Roles in explainable artificial intelligence, predictive modeling, multimodal engineering systems, and AI-assisted surgical planning were also supported by emerging evidence. Nevertheless, issues of generalizability, interpretability, regulatory aspects, and clinical application are significant obstacles to translation.

Conclusion

Artificial intelligence and deep learning demonstrate a significant potential in revolutionizing the glaucoma diagnosis and management with automated detection and progression prediction, explainable diagnostics, and decision support that is engineered. Despite the difficulties, ongoing progress in validation, explainability, and translational integration can promote the creation of precision glaucoma care in the future

pdf