A group of professionals from Moorfields Eye Hospital and the UCL Institute of Ophthalmology (IoO) designed a useful AI model. This AI model, RETFound, could have significant public health implications. The model can identify eye-related ailments. Also, it can predict various conditions. These conditions include diabetic heart attacks, strokes, eye disease, glaucoma, and Parkinson’s disease. It does this by analyzing retinal images. The NHS data was very instrumental in the system’s training.
Instead of relying on language-based algorithms like ChatGPT, RETFound harnesses imagery from millions of retinal scans obtained through the NHS. The system’s training method involves the application of a “masked auto-encoder” technique. This technique obscures approximately 80% of an image. This compels the model to deduce the concealed pixels. The developers have chosen to release RETFound as an open-source tool, allowing researchers worldwide to employ this algorithm. RETFound represents a fusion of world-class NHS data with exceptional computer science expertise from a UK-based university.
Studies that Birthed RETFound
The unveiling of RETFound follows a study led by Professor Pearse Keane’s team. This research revealed that AI-enhanced eye scans can serve as a diagnostic tool for identifying Parkinson’s disease. This research employed AI to scrutinize a dataset comprising 154,830 individuals. These individuals were aged 40 or older. Also, they received care at secondary care ophthalmic hospitals in London between 2008 and 2018.
There was a replication of the study’s process using data from the UK Biobank. It involved an assessment of 67,311 healthy volunteers aged between 40 and 69, recruited between 2006 and 2010. The study found distinctive changes in the eyes of individuals with Parkinson’s disease. There was a reduction in the thickness of the ganglion cell-inner plexiform layer (GCIPL) and inner nuclear layer (INL).
There is optimism that this technology could be further developed into a preliminary screening tool for individuals at risk of developing Parkinson’s, enabling them to take proactive measures to delay its onset.
The featured image is from NewAtlas
