Optic disc segmentation
Deep learning models trained on 180,000+ annotated fundus images identify cup-to-disc ratio, disc haemorrhages, and neuroretinal rim thinning.
Software as a Medical Device · Kraków, Poland
Early glaucoma detection,
powered by fundus AI
Retinotech develops regulated medical software for automated screening of glaucoma suspect features from color fundus photographs — helping ophthalmology networks triage faster across Central Europe.
0
Research pilot sites
0
Images analysed
94.2%
Sensitivity (validation)
Refer — Glaucoma Suspect
Optic disc: vertical C/D ratio elevated · RNFL defect pattern detected
Model confidence: 91.4%
Our platform
OpticFlow™ — decision support for glaucoma triage
Unlike general diabetic retinopathy screeners, OpticFlow™ is purpose-built for optic nerve head analysis and glaucoma suspect referral workflows in primary and community eye care.
Refer / no-refer output
Binary triage aligned with EGS guidelines. Integrates with PACS and EMR via DICOM and HL7 FHIR — no proprietary hardware required.
Explainable heatmaps
Grad-CAM visualisations highlight regions influencing the model decision, supporting clinical review and audit trails under EU AI Act requirements.
Edge & cloud deployment
Run on-premise for data residency (Poland, EU) or via our EU-hosted cloud. Typical inference time under 4 seconds per image pair.
Regulatory pathway
Preparing for market authorisation
Retinotech is developing OpticFlow™ as a software medical device and is actively preparing submissions for EU MDR clearance, US FDA premarket review, and ISO 13485 quality management certification.
- Multi-centre clinical validation completed (n=4,218)
- Technical documentation and CER in preparation
- Intended use: screening aid, not standalone diagnosis
- Anticipated classification: EU MDR Class IIa (Rule 11)
EU
MDR Class IIa
Submission in preparation
US
FDA 510(k)
Pathway under scoping
ISO
13485 QMS
Implementation underway
ISO
27001
Cloud security — planned
Testimonials
What clinicians say
"We ran an OpticFlow pilot in our community screening pathway in 2024. Referral accuracy improved and our ophthalmology backlog dropped by roughly 30% during the evaluation period."
"The heatmap overlays make MDT review straightforward. For a Polish-built SAMD still in development, the validation study design and clinical rigour impressed our research committee."
"Pilot setup took two weeks including DICOM routing. Their Kraków team provided on-site validation support — rare for AI software companies."