FDA-Approved AI Devices for Eye Care: Full List 2026
As of 2026, the FDA-authorized artificial-intelligence (AI) devices most relevant to eye care are a small, well-defined group: autonomous diabetic retinopathy detection systems — LumineticsCore, EyeArt, and AEYE-DS — plus Notal Vision’s SCANLY Home OCT for at-home retinal monitoring. Each reached the market through a specific FDA pathway, and you should always confirm current status in the FDA’s own database before relying on it.
How to read “FDA-approved” for AI eye-care devices
“FDA-approved” is used loosely in marketing, so it helps to be precise. Most AI eye-care software reaches the U.S. market by one of two routes: a De Novo authorization, used for a novel, low-to-moderate-risk device with no existing equivalent, or a 510(k) clearance, which shows the device is substantially equivalent to a legally marketed predicate. Full pre-market approval (PMA) is reserved for higher-risk devices and is uncommon for this category. The distinction matters: a De Novo authorization creates a new device type, while later 510(k) products often clear against those pioneers as predicates.
Autonomous diabetic retinopathy detection systems
The most established category is autonomous detection of diabetic retinopathy (DR) — software that reads retinal images and returns a referable-or-not result without a specialist grading the image first. These tools are designed for point-of-care use in primary care and optometric settings, and they anchor the practical side of the AI-in-eye-care story. If you are evaluating how these fit a routine diabetic workup, our guide to diabetic eye exam AI screening tools for optometrists covers the workflow in depth.
| Device | Maker | Cleared use | FDA pathway |
|---|---|---|---|
| LumineticsCore (formerly IDx-DR) | Digital Diagnostics | Autonomous detection of more-than-mild diabetic retinopathy in adults with diabetes | De Novo (2018) |
| EyeArt | Eyenuk | Autonomous detection of more-than-mild and vision-threatening diabetic retinopathy in adults | 510(k), 2020 |
| AEYE-DS | AEYE Health | Autonomous detection of referable diabetic retinopathy from retinal images | 510(k), 2022 (later handheld-camera clearance, 2024) |
LumineticsCore (formerly IDx-DR)
Made by Digital Diagnostics, this was the first autonomous AI diagnostic system the FDA authorized in the U.S., via the De Novo pathway in 2018. It is indicated to detect more-than-mild diabetic retinopathy in adults with diabetes and returns a result at the point of care without requiring a clinician to interpret the image.
EyeArt
Made by Eyenuk, EyeArt received 510(k) clearance in 2020. Its indication covers autonomous detection of more-than-mild and vision-threatening diabetic retinopathy in adults with diabetes, using retinal images captured in the office.
AEYE-DS
Made by AEYE Health, AEYE-DS was cleared via the 510(k) pathway in 2022 for autonomous detection of referable diabetic retinopathy. The company subsequently obtained clearance extending its use to a portable handheld camera in 2024, aimed at screening outside traditional imaging suites.
Home OCT monitoring with AI analysis
Beyond DR screening, the FDA granted De Novo authorization in May 2024 to Notal Vision’s SCANLY Home OCT, a patient self-operated optical coherence tomography (OCT) device. Rather than diagnosing autonomously, it lets patients capture retinal scans at home; AI-based analysis of retinal-fluid biomarkers then supports physician remote monitoring, initially for patients with neovascular (“wet”) age-related macular degeneration (AMD). It represents a different model of AI in eye care — continuous, between-visit monitoring rather than a single office screen.
What about glaucoma and other conditions?
Diabetic retinopathy and AMD monitoring are where autonomous, AI-specific authorizations are clearest today. Glaucoma is a harder problem — early disease is subtle and diagnosis blends structural and functional data — so the AI landscape there leans toward decision-support and analytics that assist, rather than replace, the clinician’s judgment. We track that separately in our review of glaucoma detection software and AI tools for optometrists. Some newer imaging hardware is FDA-cleared with AI analysis planned or in development; treat “AI-enabled” hardware and a cleared autonomous diagnostic as different things, and read each device’s actual indication.
Verify before you rely on any listing
Device names, makers, indications, and clearance status change — companies rebrand, expand indications, or are acquired. Any list, including this one, is a snapshot. Before adopting or citing a device, confirm its current authorization and exact cleared indication directly with the U.S. Food and Drug Administration. The FDA maintains searchable databases of De Novo authorizations and 510(k) clearances where you can check the current record for each product by name or maker.
What FDA authorization does and doesn’t tell you
- It confirms a specific indication — a device is cleared for a defined use and population, not for anything the marketing implies.
- It is tied to defined inputs — several DR systems are validated with particular cameras or image protocols; using unvalidated hardware falls outside the clearance.
- It does not certify performance in your clinic — reported sensitivity and specificity come from the pivotal studies, not your patient mix.
- It does not remove your responsibility — an un-gradable or borderline result is a prompt to look closer, and symptomatic patients still warrant clinical evaluation regardless of a negative autonomous read.
Choosing among cleared tools
For most optometric practices, the practical decision is between the autonomous DR systems, and the deciding factors are camera compatibility, workflow fit, reporting, and cost rather than the FDA pathway alone. Sensitivity protects against missing disease; specificity keeps false positives and needless referrals down. Our comparison guide to AI retinal screening software lays out how these tools differ in day-to-day use. Whichever you pick, confirm the cleared indication matches how you intend to use it, verify HIPAA-compliant handling of retinal images as protected health information (PHI), and re-check the device’s FDA status periodically rather than assuming a one-time authorization still stands.
