Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: Jul 26, 2025
Date Accepted: Oct 11, 2025
(closed for review but you can still tweet)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Geographic atrophy secondary to age-related macular degeneration detection and management via non-invasive retinal images and artificial intelligence: a systematic review
ABSTRACT
Background:
Geographic atrophy (GA), the endpoint of dry age-related macular degeneration (AMD), is irreversible. The recent approval by the FDA of complement C3 inhibitor, marks a significant breakthrough as the first approved treatment for GA. Therefore there is an urgent and unmet need for early detection and management of GA.
Objective:
We aimed to assess the performance of artificial intelligence (AI) using non-invasive imaging modalities and compare it with ground truth.
Methods:
We searched six electronic databases up to 1 December related to AI for GA secondary to dry AMD via noninvasive retinal images. QUADAS-AI and PROBAST were applied to evaluate the risk of bias and application.
Results:
Of 803 records identified, 155 reports were assessed for full text and 34 included GA detection (n=9), GA assessment and progression (n=16), and GA lesions prediction (n=9). The reviewed studies collectively involved at least 24,592 participants (detection: 7,019; assessment and progression: 13,251; prediction: 4,448) , with a wide age range of 50 to 94 years. The studies conducted span a diverse array of countries, including the United States, the United Kingdom, China, Australia, France, Israel, Italy, Switzerland, and Germany, as well as a multicenter study encompassing seven European nations. The studies utilized a variety of imaging modalities to assess GA, including CFP, FAF, NIR, SD-OCT, SS-OCT, and 3D-OCT. DL algorithms (e.g., U-Net, ResNet50, EfficientNetB4, Xception, Inception v3, PSC-UNet) consistently showed remarkable performance in GA detection and management tasks, with several studies achieving performance comparable to clinical experts.
Conclusions:
AI, especially DL-based algorithms, holds considerable promise for the detection and management of GA secondary to dry AMD with performance comparable to ophthalmologists. It has strong potential to replace the traditional technologies in clinical settings, however, further research is needed to robustly enhance their reporting specification, data diversity and implement rigorous external validation. Clinical Trial: This systematic review was registered with PROSPERO (CRD420251000963).
Citation
