, heralded as one of the earliest foundation models in healthcare and the very first in ophthalmology. Built on a vast dataset of eye scans from the NHS, RETFound stands as an open-source beacon, freely accessible to institutions worldwide. It represents a monumental leap forward in the global endeavor to employ
in the fight against blindness.
In a world where AI’s potential is racing ahead at an astonishing pace, foundation models like RETFound are leading the charge. Unlike conventional AI systems, RETFound not only outperforms existing models across various clinical tasks but also excels in diverse populations and rare disease cases. This monumental achievement has the potential to revolutionize eye examinations for the 21st century.
Professor Pearse Keane, a senior author of this research, envisions a future where the UK, armed with unparalleled clinical data from the NHS and cutting-edge computer science expertise from its universities, emerges as a global leader in AI-powered healthcare. He sees their work as a blueprint for this transformative journey.
Advertisement
A recent report by the UK government hailed AI foundation models as a “transformative technology.” In this vein, RETFound stands as a testament to the immense possibilities when such models are applied to eye scans. By leveraging an innovative self-supervising approach, RETFound achieves remarkable label efficiency, requiring as little as 10% of human labels in its dataset (1✔ ✔Trusted Source
A foundation model for generalizable disease detection from retinal images
Go to source
).
Eyes: A Window to Overall Health
Beyond eye diseases, RETFound has the potential to revolutionize our understanding of general health through the lens of ‘oculomics’ – a term coined by Professor Alastair Denniston in 2020. The eyes offer a unique window into our overall health, unveiling the secrets of our nervous system. This knowledge is crucial in tackling complex diseases and the challenges associated with aging.
Furthermore, RETFound’s effectiveness extends across diverse populations, ensuring its impact reaches far and wide. Lead author Yukun Zhou emphasizes how their model, trained on datasets reflecting the rich ethnic diversity of London, lays a strong foundation for researchers worldwide to advance healthcare applications, from ocular disease diagnosis to systemic disease prediction.
The collaboration between Moorfields Eye Hospital and the UCL Institute of Ophthalmology has not only birthed an AI system of unparalleled potential but also ignited a beacon of hope for millions around the globe.
“The power AI lies not just in its ability to detect eye diseases, but in its promise to transform how we approach healthcare and envision a future where our eyes reveal more than we ever thought possible.
Reference :
- A foundation model for generalizable disease detection from retinal images – (https:www.nature.com/articles/s41586-023-06555-x)
Source: Medindia
