AI-generated conceptual illustration; screen images are illustrative and contain no patient data.
A person can read an eye chart well and still have changes developing in the retina or optic nerve. Modern imaging helps clinicians examine these structures in more detail. Optical coherence tomography, or OCT, is now an established part of many eye-care pathways. Artificial intelligence is also used in some carefully defined screening services.
These technologies answer different questions. Understanding what each test measures can make a scan report much less confusing and help patients know when further examination is needed.
OCT looks beneath the surface
A retinal photograph shows the back of the eye from the front. OCT uses reflected light to produce cross-sectional images, revealing layers within the retina. It does not use X-rays. The technique was first described in a landmark 1991 research paper and has developed into a widely used clinical tool. Huang and colleagues, Science
OCT can help an ophthalmologist identify retinal thickening or fluid, assess the macula and measure structures relevant to glaucoma. In macular disease, repeated scans help show whether an abnormality is changing and can inform treatment decisions. The National Eye Institute describes OCT as an additional test that may be used alongside a dilated examination when assessing age-related macular degeneration. National Eye Institute: AMD
For the patient, scanning usually means resting the head on a chinrest and looking at a target. Eye drops to widen the pupils may be needed depending on the examination and image quality. The clinician combines the result with symptoms, vision testing and examination findings rather than treating the scan as a diagnosis by itself.
OCT angiography adds a view of blood flow
OCT angiography, often written OCTA, compares repeated scans to identify motion from blood cells. This creates maps of small blood vessels without an injected contrast dye. It can provide useful information about retinal and nearby vascular networks.
However, an attractive vessel map is not automatically an accurate one. Eye movement, poor focus and software errors in separating tissue layers can create misleading patterns. In a 2019 study of 75 eyes, investigators found several kinds of artifacts and emphasized their effect on interpretation. The study illustrates why clinicians review the underlying images and quality, not just the processed map. Enders and colleagues, 2019
Other research in more than 4,000 people found artifacts in both glaucoma and normal eyes. A device’s quality score alone did not reliably identify every problem. A scan may therefore need repeating or comparison with another test before it can guide care. OCTA artifact study, 2022
Where AI screening has a proven role
Some medical AI systems are designed to analyze retinal photographs for a specific condition, such as diabetic retinopathy. A pivotal 2018 trial tested one autonomous system in primary-care settings among adults with diabetes and no known diabetic retinopathy. It met pre-specified performance targets against a reference assessment, supporting a defined clinical use. Abràmoff and colleagues, 2018
That finding does not validate every eye-health app or every algorithm. Performance depends on the particular system, image acquisition and the population in which it is used. An AI tool authorized for diabetic retinal screening should not be assumed to diagnose glaucoma, cataract or every cause of blurred vision.
The FDA’s original authorization documentation for IDx-DR explicitly limits its purpose and describes what to do when image quality is insufficient or referral is recommended. An ungradable image is not a reassuring normal result. FDA device review
What to ask about your test
Useful questions include: What question is this scan meant to answer? Was the image quality adequate? Has anything changed since the last scan? Is AI being used, and for which condition? Who will arrange follow-up if the result needs attention?
A comprehensive eye examination remains important when symptoms or risk factors need assessment. Sudden loss of vision, a new curtain or shadow, or a sudden increase in flashes and floaters requires urgent medical assessment, even after a recent reassuring screening result. Imaging is most useful when the result leads to an appropriate clinical decision and timely follow-up.
Related reading
Diabetic retinopathy | Glaucoma
This article provides general education and cannot determine which treatment is right for an individual. Diagnosis and treatment require an ophthalmologist’s assessment. Technologies, approvals and availability vary by country and centre.
Sources
- Huang D et al. Optical coherence tomography. Science 254:1178–1181. (1991-11-22).
- National Eye Institute. Age-Related Macular Degeneration. (Accessed 2026-10-05).
- Enders C et al. Quantity and quality of image artifacts in optical coherence tomography angiography. PLoS ONE 14:e0210505. (2019-01-25).
- Assessment of Artifacts in Swept-Source Optical Coherence Tomography Angiography for Glaucomatous and Normal Eyes. Translational Vision Science & Technology. (2022).
- Abràmoff MD et al. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. npj Digital Medicine 1:39. (2018-08-28).
- U.S. FDA. De Novo Classification Request for IDx-DR. DEN180001. (2018).
Evidence checked: 5 October 2026.
