Ophthalmic foundation model

The intelligencelayer ofeye care.

One foundation model for every image, every device and every visit.

FUNDUS · OCT · ANGIOGRAPHY · ANTERIOR SEGMENT · TOPOGRAPHY · RECORD

Demand is rising. Capacity is shrinking.Disease is hidden.

Demand

60M†

Americans at risk for retinal disease.

Capacity

−30%1

Projected US ophthalmology workforce gap by 2035.

Hidden

50%†

Of people with an irreversible eye disease don't know they have it.

Time

>50%2

Of an ophthalmologist's time per patient is spent in the record, not with the patient.

Missed

>25%3

Of clinically significant cases are missed at routine eye exams.

Patientdataisscatteredacrossdevicesthatdon'ttalktoeachother,andscarcespecialisttimeisspentonrecordsandonpatientswhodidn'tneedtobethere.

Platform

Integrate.Then augment.
FundusOCTOCT-AVisual fieldsFAFSlit lampEMROphthalmicFoundationmodelScreening & triageDisease detectionReferral prioritizationProgression forecast

01 · Integrate

Integrate

Every ophthalmic image and the medical record, together in one view.

02 · Augment

Augment

Physician-in-the-loop decision support from one multimodal model.

The physician always makes the final call.

One foundation model. Multiple modalities.Multiple diseases.

A device-agnostic intelligence layer that powers multiple clinical products.

Fovea™

Integrated view

Every image and record in one place.

  • Vendor-neutral: native data from any device, not screenshots
  • One timeline: scans, pressures, medications, diagnoses
  • Printouts become trendable data
Colour fundus photograph of a retina

Fundus

Greyscale OCT B-scan of a retina

OCT

Greyscale visual field printout

Visual field

Intraocular pressure18 mmHg

Insight™

Pan-ophthalmic decision support

Multi-disease answers from one image.

  • Fundus and volumetric OCT, combined
  • Attention maps show where the model looks
  • Lesion-level detail, high specificity on healthy eyes
Fundus photograph with model attention overlay
Diabetic retinopathy · ICDR grade 2DME risk: High

In the clinic

One photo. Three diseases.Inside a visit the patient already attends.
No dilationNo eye doctor on siteInside an existing visit

Results

See more. Miss less.Refer smarter.
−77%4

Missed referable disease vs. ophthalmologists

Benchmarked against 10 ophthalmologists from 8 centers, on glaucoma, AMD and diabetic retinopathy. No referable case was caught by the readers and missed by the model.

−60%5

False positives vs. non-specialist graders

Referable diabetic retinopathy.

One photo, three diseases.

6

Diabetic retinopathy

Sensitivity
93%
Specificity
94%

AMD

Sensitivity
92%
Specificity
90%

Glaucoma

Sensitivity
94%
Specificity
99%

Every 100 diabetics screened

Today

100

eye-doctor visits needed, and disease is missed when patients skip them.

With Lumiere

15

eye-doctor visits, and 9+ of 10 diabetic retinopathy cases caught.

~85 cleared on the spot~15 referred<1 case missed

Based on published validation. Assumed prevalence: DR 10%, AMD 12%, glaucoma 5%.

From screening to prioritization.

252 patients, a portable handheld camera the model had never seen. No fine-tuning. Image only.

0.9897

AUC · Referable DR, zero-shot

11vs1287

Unnecessary referrals vs. a clinical-data model

98%7

Vision-threatening cases seen within 4 weeks

Vision-threatening DR seen within 4 weeks · demand 25% over capacity

Lumiere ranked queue
98%
Clinical triage
78%
First come, first served
34%

Ranked as Very urgent · Urgent · Priority · Routine. Note: manuscript submitted.

Unnecessary referrals measured at matched 97% sensitivity.

Science

Peer-reviewed. Indexed.Open.

The models behind Lumiere are built on published, peer-reviewed research in glaucoma, retina, OCT and retinal vasculature.

16

Peer-reviewed papers

5

Research areas

2026 · Int J Retina Vitreous↗

Artificial intelligence in retinal care: transforming the doctor-patient partnership

How AI is reshaping the relationship between retina specialist and patient, and how to deploy it responsibly.

2026 · npj Digital Medicine↗

Shifting the retinal foundation models paradigm from slices to volumes for optical coherence tomography

Reading the full 3D OCT volume instead of a single slice: average AUROC 0.94 vs 0.90 for the best image-based model.

2026 · Physiological Measurement↗

Ophthalmology foundation models for clinically significant age macular degeneration detection

A benchmark of six foundation models on 70,000 expert-labelled fundus images from seven datasets, for moderate-to-late AMD.

2026 · IEEE Transactions on Biomedical Engineering↗

GONet: a generalizable deep learning model for glaucoma detection

Trained on more than 119,000 fundus photographs from seven datasets; out-of-distribution AUC 0.88–0.99.

2025 · Ophthalmology Science↗

Computerized analysis of the eye vasculature in a mass dataset of digital fundus images: the example of age, sex, and primary open-angle glaucoma

An automated vasculature pipeline applied to 32,768 fundus images; glaucoma shows a pattern consistent with accelerated vascular ageing.

2025 · Physiological Measurement↗

Deep learning generalization for diabetic retinopathy staging from fundus images

DR staging developed on 91,984 images from six independent datasets, designed to generalize across populations.

2024 · Physiological Measurement↗

LUNet: deep learning for the segmentation of arterioles and venules in high resolution fundus images

High-resolution artery/vein segmentation that outperformed three benchmark methods on a local test set and four external test sets.

2024 · Scientific Data↗

Leuven-Haifa high-resolution fundus image dataset for retinal blood vessel segmentation and glaucoma diagnosis

An open dataset of 240 expert-corrected artery/vein annotations with glaucoma labels.

2023 · Computer Methods and Programs in Biomedicine↗

FundusQ-Net: a regression quality assessment deep learning algorithm for fundus images quality grading

Automated 1–10 image-quality grading; 99% accuracy on an external test set. It is the quality gate in front of every analysis.

Selected publications · 9 of 16

Live demo

One photograph.A full reading.

Pick a sample case and run the model. Insight™ reads the image, shows where it is looking and drafts the finding. The physician still makes the call.

Ready
Fundus photograph of a left eye with diabetic retinopathy
Fundus · Left Eye
Image quality
Diabetic retinopathy
DME risk
Age-related macular degeneration
Glaucoma

Demonstration on sample data. Not a medical device and not for clinical use.

Whattheeyereveals,weilluminate.

Contact

Let’s talk.

Clinical collaborations, deployment and integration. Tell us what you’re working on.

contact@lumiere-ai.com

References & notes

  1. 1US workforce gap: published ophthalmology workforce projections for the United States, 2020–2035 (Ophthalmology, 2024).
  2. 2Time in the record: EHR audit-log study in an academic ophthalmology center (JAMA Ophthalmology, 2017); >50% calculated from 10.8 min EHR use of ~19 min total physician time per encounter.
  3. 3Missed cases at routine exams: JAMA Ophthalmology, 2017; and a 2023 study on missed glaucoma diagnosis by eye-care providers.
  4. 4Expert-reader benchmark: 91 fundus images, 55 referable; Machine Learning for Health, 2026.
  5. 5False-positive comparison: Lumiere specificity 94.0% vs 84.8% for trained non-specialist graders using retinal photographs (Diabetic Medicine, 1998).
  6. 6Per-disease sensitivity/specificity: published Lumiere validation studies.
  7. 7Handheld zero-shot study: 252 patients, 415 eyes; manuscript submitted.
  8. †Figures marked † are population-level estimates from published epidemiology, not from Lumiere studies.

Lumiere products are investigational and not yet cleared by the FDA. Performance figures come from published and submitted validation studies.

Lumiere.ai

Smarter eye care. Better health intelligence.

Lumiere
© 2026 Lumiere.aiFor investigational use. Not cleared by the FDA.