RetiVista is an AI-assisted screening tool in development: it reads a photograph of the retina and flags the signs of diabetic retinopathy, so that people with diabetes can be referred to an eye doctor before sight is lost.
How it works Get in touch
Diabetes damages the small blood vessels of the retina, the light-sensitive layer at the back of the eye. This damage — diabetic retinopathy — is silent for years and is one of the leading causes of blindness in working-age adults. Yet most of that vision loss can be avoided when it is found early and treated.
The problem is not the treatment: it is getting people to the eye exam. Many people with diabetes skip their yearly retinal check because it means a specialist, a waiting list, a trip.
RetiVista is designed to bring the first look closer to the patient: a retinal photograph taken by a pharmacist, a nurse or a family doctor, analysed on the spot by software that estimates the severity of retinopathy and says whether the person should be referred to an ophthalmologist.
A colour picture of the back of the eye, taken with a portable retinal camera. No eye drops, no dilation, no discomfort: a few seconds per eye, in a pharmacy, a clinic or a screening van.
The software examines the image and estimates the severity of diabetic retinopathy on the international clinical scale (0 = none … 4 = proliferative). It runs on an ordinary computer, on site: no image needs to leave the room.
The output is simple: referable — see an eye specialist — or not referable — repeat the check next year. It is a screening signal, not a diagnosis: the ophthalmologist always has the last word.
Automatic reading of retinal photographs for diabetic retinopathy is not science fiction: it is a mature field, and since 2018 regulators have authorised autonomous screening software of this kind. What makes the difference is not the algorithm alone, but how carefully it is validated on real patients and how it fits into everyday care.
The RetiVista prototype is trained on public collections of retinal photographs graded by specialists, and it is measured only on images it has never seen — including a collection from a different country and a different camera, which is the honest test of whether a model generalises. Results are published in the project notes as the work progresses.
Thousands of retinal photographs graded by ophthalmologists on the international 0–4 scale.
Accuracy is reported only on held-out images and on an independent collection from another country.
Analysis runs locally. Images stay where they are taken; nothing is uploaded to a cloud.
A quick retinal check where people already collect their medicines or see their family doctor, with a clear referral when it matters.
A way to make sure every patient gets the yearly retinal check, and that the ones who need the specialist are seen first.
Population screening in towns and workplaces, with portable cameras and immediate triage — fewer unnecessary specialist visits, fewer missed ones.
Model trained on public collections; local demo application; honest measurements on unseen images.
Real photographs with confirmed diagnoses, together with a diabetes and eye-care centre in Tuscany. Portable camera in the field.
Medical-device quality system, clinical evaluation, CE marking under the EU Medical Device Regulation; FDA pathway for the United States.
Screening as a service for pharmacies, clinics and programmes — with the data and the partnerships that make it trustworthy.
RetiVista is developed by Zannini Group, the holding of the Farnesi family, through its Italian company Zannini Gemini S.r.l. of Pisa — founded in 1991, with experience in electromedical devices.
The project is led by Bruno Farnesi, an entrepreneur and former Italian Navy helicopter pilot, with a career spanning pharmaceutical packaging, shipbuilding and medical devices. The approach is the family's: build things properly, one step at a time, and say clearly what is proven and what is not yet.
“The value is not the algorithm. It is the care with which it is validated, and the people it helps reach an eye doctor in time.”
We are looking for partners for the clinical pilot: diabetes centres, ophthalmology departments, pharmacy networks, and makers of portable retinal cameras. If this is you, write to us.
Contact