We are pleased to announce the publication of a new AI-AGE research paper in Current Medical Imaging: “Deep Learning Discrimination of Colorectal Cancer using Retinal Fundus Photographs: UK Biobank Case-Control Study,” by Ivan Jovovic, Tomo Popovic, Isidora Rubezic Lukic, Dejan Babic, and Natasa Popovic.
The study investigates whether deep learning analysis of retinal fundus photographs can distinguish individuals with colorectal cancer (CRC) from those without CRC. Using 1,600 retinal images from the UK Biobank and an EfficientNet-B3 deep learning model, the study achieved an AUROC of 0.945 on an independent participant-level test set.
The findings support the hypothesis that retinal images may contain measurable signatures associated with systemic diseases beyond ophthalmology. At the same time, the study shows that retinal AI should not currently be considered a stand-alone diagnostic tool for colorectal cancer. Its potential value is primarily as a future non-invasive triage or risk-enrichment approach, subject to further validation in longitudinal and clinically representative populations.
This publication represents an important output of the AI-AGE project and directly contributes to its research objectives related to the use of artificial intelligence and retinal imaging for identifying novel non-invasive biomarkers associated with aging and age-related diseases.
The article was published online on 11 September 2026 in Current Medical Imaging and is currently available as an Early View article [link].