Ophthalmology · Ophthalmic image/data-management platform with AI-assisted retinal analysis
RetinAI Discovery
RetinAI (Ikerian AG, an EssilorLuxottica company)
RetinAI Discovery is best understood not as a single diagnostic device but as an ophthalmic data platform with AI analysis bolted on — and its regulatory story turns entirely on that distinction. In the United States, the FDA clearance covers the image- and data-management platform; the AI modules that do the interesting work, retinal-layer and fluid segmentation and biomarker quantification, are research-use-only. In Europe those same modules carry a CE mark under the EU Medical Device Regulation. The evidence is a genuine mix: an independent academic study from Bern used Discovery's segmentation to characterise inherited retinal disease with roughly 91% accuracy, while the strongest diabetic-retinopathy numbers come from a manufacturer-linked multicentre study of the LuxIA screening algorithm whose headline figures are company-reported. The platform is deployed heavily in pharmaceutical and clinical-trial workflows rather than as a front-line autonomous screener, and no outcome trial establishes that its use changes patient care. A clean public safety record and a real independent validation are worth a mark; the research-use-only US status of the AI, the single EU marketing pathway for the clinical AI, and the absence of outcome evidence keep it to one.
Performance Metrics
Clinical Evidence
Discovery's evidence base reflects what the platform actually is — an analysis and data layer used across research and clinical workflows — and it should be read with the US research-use-only status of the AI modules kept firmly in view. The clearest independent evidence is an academic study from the University of Bern / ARTORG and Inselspital (Translational Vision Science & Technology 2025; PMID 41342623), which used Discovery's OCT segmentation to characterise inherited retinal diseases. Across 327 OCT images from 181 patients with inherited retinal disease plus 146 controls, automated segmentation of six retinal layers and detection of nine biomarkers supported detection of inherited retinal disease with approximately 91% accuracy. The authors are independent academics using Discovery as a tool, which makes this the most credible external signal of the segmentation module's utility. The strongest diabetic-retinopathy result is manufacturer-linked and should be labelled as such. The CARDS study (BMJ Open Ophthalmology 2025; PMID 40340790) validated the LuxIA algorithm — a single 45-degree colour-fundus screen run on the Discovery platform — for more-than-mild diabetic retinopathy across five Spanish university hospitals, with images collected December 2021 to December 2022. RetinAI reports a sensitivity of 97.1% and specificity of 94.8%; those specific figures and the study's roughly 945-patient sample are company-reported and were not independently re-verified against the primary record on this review, so they are presented as manufacturer-reported rather than confirmed. The wider literature around Discovery is largely validation and real-world use rather than autonomous-diagnosis trials, and several nAMD fluid-quantification papers are manufacturer-linked with sample sizes that vary across sources; those are not featured here pending confirmation. No randomised or outcome trial of Discovery was located. The platform's demonstrated value is measurement, standardisation and data organisation for clinicians and researchers — not autonomous reading.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Peter VG, Hayoz M, Scandella D, et al. (independent; University of Bern / ARTORG / Inselspital) | RetrospectiveRetrospective | 327 | — | — | ~91% accuracy for inherited-retinal-disease detection (6-layer + 9-biomarker segmentation) | Transl Vis Sci Technol, 2025; PMID 41342623 — 327 OCT images / 181 IRD patients + 146 controls |
| CARDS study group (RetinAI-linked; five Spanish university hospitals) | RetrospectiveRetrospective | 945 | 97.1% | 94.8% | mtmDR via LuxIA single-image screen — figures manufacturer-reported, not independently re-verified | BMJ Open Ophthalmol, 2025; PMID 40340790 |
Clinical Pulse
Verified clinician sentiment
Verified clinician reviews launching soon · Apply to contribute →
Reviews are signed by clinicians verified against their regulator's public register.
Inside the algorithm
Editorial featureHow Discovery turns multimodal retinal imaging into structured measurement.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
Images from many devices arrive on one platform.
RetinAI Discovery ingests ophthalmic imaging from a wide range of devices — OCT, OCT-angiography and colour fundus photography — into a single cloud, web-based platform. It stores, displays and lets clinicians and researchers compare studies across visits, and supports grading surveys and electronic case report forms for clinical research.
This data-management layer is the part cleared by the FDA in the United States; the analysis that follows is where the jurisdictional scope narrows.
Input
Multimodal ophthalmic imaging
Inference
AI modules — CE-marked in EU, RUO in US
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
K211715
Class II
Safety Record
No RetinAI or Ikerian device recalls, FDA safety communications or MAUDE adverse-event reports were identified in publicly available sources as of July 2026; because the FDA MAUDE and enforcement databases could not be queried directly from this environment, this should be read as "none found in public reporting" rather than an exhaustive audit. The residual risk profile is bounded by what the software does: it is an assistive analysis and data-management layer, not an autonomous reader, and in the United States its AI analysis modules are research-use-only, so they are not to be relied upon for US clinical decisions. The principal caution is the general one for quantitative decision-support — a segmentation or biomarker output supports, but does not replace, the clinician's interpretation of the underlying images.
Intended Use & Indications
RetinAI Discovery is a cloud, web-based platform for managing and analysing ophthalmic imaging across clinical and clinical-research settings. It imports, stores, displays and compares multimodal images — optical coherence tomography, OCT-angiography and colour fundus photography — from a wide range of ophthalmic devices, and supports side-by-side comparison across visits, grading surveys and electronic case report forms for research. On top of this data layer sit AI-assisted analysis modules: automated segmentation of retinal layers and of intraretinal fluid, subretinal fluid and pigment-epithelial detachment, quantification of OCT biomarkers, and LuxIA, a cloud algorithm that screens for more-than-mild diabetic retinopathy from a single 45-degree colour fundus image. The regulatory scope differs sharply by jurisdiction and is the most important thing to hold in view: in the United States the FDA 510(k) clearance covers the image/data-management platform, and the AI analysis modules (layer and fluid segmentation, biomarkers, advanced segmentation and geographic-atrophy modules) are labelled research-use-only; in Europe the platform and three AI models carry an EU MDR CE mark. In every setting the software is assistive: it produces quantitative measurements and organised views for the clinician's decision, and does not make an autonomous diagnosis.