The retinal fundus photograph is one of medicine's most information-dense, least exploited investigations. In under five minutes, a non-invasive snapshot of the retina can reveal not only ocular disease but systemic conditions — including hypertension, diabetes, and early cardiovascular risk — years before clinical symptoms emerge.

EyeMap.ai Labs was founded to close the gap between what retinal imaging can show and what current clinical workflows actually capture. We train, validate, and deploy deep learning models that operate on standard fundus images, making population-level screening feasible at point-of-care settings including optometry practices.

Research Principles

Our work is guided by three commitments: clinical validity (models are benchmarked against expert ophthalmologist grading), reproducibility (we document training data, architectures, and evaluation protocols), and equity (we test model performance across demographic subgroups to detect and mitigate bias).

Oculomics

OCULOMICS is the study of the eye as an indicator of overall health, coined in 2020 by Prof. Denniston. Our programme extends this framework by using retinal imaging biomarkers to investigate cardiometabolic and neurovascular risk patterns that may be detectable before overt clinical disease.

Regulatory Philosophy

All models are developed with the EU AI Act and MDR regulatory pathway in mind. We do not present AI outputs as diagnostic conclusions — they are screening flags that direct clinician attention. Human oversight is central to every deployment design.

UK Biobank Application

Population Validation Study

  • Accessing retinal imaging from UK Biobank cohort
  • Validating AMD, DR, and glaucoma models at scale
  • Longitudinal biomarker analysis
  • Subgroup performance assessment
Active Models

Diagnostic AI Systems

  • Age-Related Macular Degeneration (5-class)
  • Diabetic Retinopathy (5-class, ETDRS-aligned)
  • Glaucoma via Cup-to-Disc Ratio estimation
Ethics & Governance

Oversight Framework

  • GDPR-compliant data processing
  • Institutional ethics approval (pending)
  • UK Biobank governance compliance
  • Participant rights documentation

Diagnostic model overview


Prospective partner clinics

Partnership discussions are underway with multiple clinics to support prospective clinical validation of retinal AI screening workflows and dataset expansion.

Retina Clinic, Bucharest
In Discussion

Working with Dr. Florian Balta to validate retinal pathology detection models in a specialized ophthalmology setting.

Spitalul Universitar de Urgență București
In Discussion

Prospective collaboration to evaluate stroke-related screening workflows in emergency-care contexts and diverse patient populations.

Vitreum
In Discussion

Partnership discussions ongoing to expand clinical validation coverage and access to additional patient cohorts.

Arcadia
In Discussion

Partnership discussions ongoing to expand clinical validation coverage and access to additional patient cohorts.

Clinic Dr. Holhoș
In Discussion

Partnership discussions ongoing to expand clinical validation coverage and access to additional patient cohorts.

Clinic Dr. Mihnea Munteanu
In Discussion

Partnership discussions ongoing to expand clinical validation coverage and access to additional patient cohorts.