Eye disease detection

The problem
Age-related eye conditions, such as macular degeneration, are easier to treat the earlier they are detected. For that, an ophthalmologist reviews eye images one by one. The question of this research: can a model flag abnormalities in those images, so the ophthalmologist looks earlier and more precisely?

Division of roles

• OMC Zaandam, an eye clinic in Zaandam and Hoorn: the eye images and the medical analysis.
• MagnaVersum: the data platform on which the images were collected and made available.
• Twentynext: the AI model, deep learning and image classification, trained on those images.

What the model does
It classifies eye images for signs of age-related eye disease at an early stage. It provides a signal the ophthalmologist takes into the assessment; it does not make a diagnosis.

What it is for

• flagging abnormalities earlier, before they stand out to the naked eye;
• supporting the ophthalmologist in diagnosis and treatment choice.

The gain is in the timing: the earlier a condition shows up, the more treatment options remain. The research was carried out with support from the Dutch MIT scheme for R&D collaboration between SMEs.

Related: medical diagnostics with AI.