Computational Ophthalmology

Computational ophthalmology uses AI to bring specialist eye care to more patients — from retinal photographs and OCT to images of the eyelid taken with a phone.

With the Department of Ophthalmology at the University Hospital Bonn we develop AI for eyelid tumours: an app that supports classifying eyelid tumours as benign or malignant, and INTERACT-CMIL, which grades conjunctival melanocytic lesions from histopathology in an interpretable way. Affordable Retina diagnoses inherited and common retinal diseases from affordable colour fundus photography, and with ZEISS we learned disentangled, fair and discriminative representations for ophthalmic imaging. GCA-US-AI diagnoses giant cell arteritis — a cause of sudden vision loss — from ultrasound of the temporal arteries.

Our results have been presented at the German Ophthalmological Society (DOG) congress and focus on fairness, interpretability and deployment at the point of care.