aceAI image generation technology in ophthalmology: Use, misuse and future applications
BACKGROUND: AI-powered image generation technology holds the potential to reshape medical practice, yet it remains an unfamiliar technology for both medical researchers and clinicians alike. Given the adoption of this technology relies on clinician understanding and acceptance, we sought to demystify its use in ophthalmology. To this end, we present a literature review on image generation technology in ophthalmology, examining both its theoretical applications and future role in clinical practice. METHODS: First…
"CRE-003 OphaGo - Retinal Imaging Adapter - Huced Despro ITS" by Djoko Kuswanto is marked with CC0 1.0. To view the terms, visit https://creativecommons.org/publicdomain/zero/1.0/.
This peer-reviewed review from March 2025 demystifies AI-powered image generation for ophthalmology, surveying literature prior to September 2024 on GANs, autoencoders, and diffusion models and cataloging reported uses such as improving diagnostic models, inter-modality transformation, prognostication, denoising, and education.
It matters because ophthalmology is highly image-dependent, so synthetic images could expand training data and clinical tools, but the source itself stresses the field is still in its infancy with unresolved issues around bias, patient data security, explainability, validation consistency, and potential misuse that must be addressed before routine clinical adoption.
- Review surveyed ophthalmology literature prior to September 2024 covering GANs, autoencoders, and diffusion models.
- Reported clinical applications include inter-modality image transformation, image denoising, and individualized education.
- Authors note future emphasis on clinically grounded metrics, foundation models for generation, and methods to ensure data provenance.
Literature prior to September 2024 reports generative models being applied in ophthalmology to improve diagnostic AI, enable inter-modality transformation, improve treatment and prognostication, denoise images, and support individualized education.
The rundown
The review first outlines model designs for synthesis including generative adversarial networks, autoencoders, and diffusion models, then surveys ophthalmology literature before September 2024 by model type and clinical application.
It identifies barriers to adoption such as computational and logistical barriers to development, challenges with model explainability, and inconsistent validation metrics, while pointing to future work on foundation models and data provenance.
Sources
- Peer-reviewedProgress in Retinal and Eye Research2025-03-17
How should this claim be treated?
The debate