AI in healthcare is often surrounded by hype. At the Eye Hospital Rotterdam, however, it is being applied in very practical ways that improve diagnostics, save time, and support collaboration between medical and technical teams.
In a recent vodcast, Vincent Bourgonje, R&D Manager at the Eye Hospital, shared how his team is using AI across ophthalmology research and patient care — from glaucoma screening to corneal transplant analysis.
Watch the full interview below:
Glaucoma causes optic nerve damage, often without early symptoms. By the time patients notice vision loss, much of the damage is irreversible.
Traditionally, only glaucoma specialists could reliably interpret fundus photos. Vincent and his colleagues are developing an AI tool that helps optometrists or even general practitioners screen for glaucoma early. The AI model, trained on more than 100,000 images, can recommend which patients need referral to a specialist.
This approach could enable population-level screening and earlier treatment.
AI models in ophthalmology need to work across diverse patient groups. A tool trained only on one population (e.g. Asian eyes) may not perform well in Europe.
To address this, the Eye Hospital validates its glaucoma models on local data from the Rotterdam Study, ensuring the tool works for real patients in its region. Current results show 74% sensitivity at 95% specificity, with ongoing efforts to improve.
For corneal transplants, it is critical to measure endothelial cell density. Traditionally, this is done manually — taking up to 15 minutes per patient.
Vincent’s team trained an AI model to count cells automatically in just seconds, improving both speed and reliability. While still in R&D, the tool is already being shared with partner institutions for validation on their datasets.
“With AI, what used to take me 15 minutes can now be done in 10 seconds.”
Vincent emphasized that success in AI depends on collaboration:
- Partnering with other hospitals to share datasets.
- Working with engineers at Erasmus MC to bring technical expertise into clinical R&D.
- Running grand challenges where different groups test models on the same data to find the best approach.
The R&D team is naturally enthusiastic, but in the clinic, adoption requires buy-in from doctors. Vincent highlighted the importance of:
- Involving clinicians early.
- Starting with small, concrete experiments.
- Letting doctors see time savings and reliability improvements directly.
Without clinical enthusiasm, even the best tool won’t make it into practice.
Some of the Eye Hospital’s AI tools face regulatory hurdles before they can be used in daily care. For example, the corneal cell counting model is not yet CE-marked, which prevents clinical use despite its proven accuracy.
Funding and commercial viability also remain challenges, especially for niche AI applications where no company is positioned to take the product to market.
Vincent believes one of the most impactful future applications will be AI-driven summarization of patient data. Ophthalmologists often spend half of a consultation reviewing notes, scans, and historical images. AI that can bring this information together could:
- Save doctors up to 50% of consultation time.
- Give patients more face-to-face attention.
- Improve decision-making by presenting the right data at the right moment.
“Do not fear, do not hype. AI is a tool. If it improves care, you will notice it quickly.”
The work at the Eye Hospital Rotterdam shows that AI in healthcare is not just theory — it’s already producing practical tools that save time, improve reliability, and enable earlier treatment.
From glaucoma screening to corneal transplant analysis, Vincent’s message is clear: collaboration, validation, and clinical engagement are the keys to making AI work in practice.
Join our live talk show on AI-Ready Operations
Discover how leaders in healthcare, pharma, and biotech are breaking down data silos, overcoming adoption barriers, and building the foundation for precision medicine. Hear from experts at Alphega Europe, Julius Clinical, and Digiteum as they share practical strategies for scaling AI beyond pilots.
REGISTER FOR THE EVENT
