Nilanjan
AI-driven technology has proved a blessing for human beings, especially in the field of medical science and technology. Researchers at UCL Institute of Ophthalmology and Moorfields Eye Hospital have developed an AI-guided system that can accurately detect the most debilitating eye diseases and also predict heart attacks, Parkinson’s disease and stroke.
RETFound is the first AI foundation model in ophthalmology and one of the first types in healthcare system. The mode is based on millions of eye scans that the research team sourced from the NHS. The research team has decided to make the system available for free to any institute worldwide. This AI-supported eye care and detection system is believed to become a keystone for global efforts to detect sight-threatening diseases and treat them. This ground-breaking invention has been published in Nature.
The field of Artificial Intelligence has undergone massive progress and development at a rapid pace. ChatGPT, one of the most talked about AI model, has already created a buzz. A foundation model refers to a robust, intricate AI system, based on a huge volume of unlabelled data that can be fine-tuned for a wide range of subsequent tasks.
RETFound has all the makings to outsmart the existing AI system across the diverse range of intricate clinical works. Moreover, the advanced state-of-the-art detection model can address a significant number of drawbacks in the existing AI system by functioning well in diverse population as well as in patients suffering from rare diseases.
Senior author Professor Pearse Keane (UCL Institute of Ophthalmology and Moorfields Eye Hospital) has said, “This is another big step towards using AI to reinvent the eye examination for the 21st century, both in the U.K. and globally. We show several exemplar conditions where RETFound can be used, but it has the potential to be developed further for hundreds of other sight-threatening eye diseases that we haven’t yet explored.”
“If the U.K. can combine high quality clinical data from the NHS, with top computer science expertise from its universities, it has the true potential to be a world leader in AI-enabled health care. We believe that our work provides a template for how this can be done”, he sounds hopeful.
The UK government described AI foundation models as “a transformative technology” in an early 2023 report. Since ChatGPT was launched in November, 2022, the AI technology has become more popular and much talked-about across the spectrum of medical science and other fields as well. Even common people are curious to know about its use, importance and future in different fields.
ChatGPT was a foundation model developed using a large quantity of text data to program a versatile language tool.
The research team adopted a similar approach while developing the world’s first RETFound trained on millions of retinal scans to develop a model adaptable for potentially limitless uses.
One of the most important challenges when creating AI models is the need for efficient human labels which are time-consuming and costly to acquire. According to what is demonstrated in the paper, RETFound is capable of matching the performance of other AI systems despite using only 10% of human labels in its dataset.
The remarkable improvement in data efficiency has been possible by adopting an innovative self-supervising approach that involves RETFound in masking parts of an image and learning to predict the missing parts itself.
RETFound could be a ground-breaking development in detecting some of the most sight-threatening eye diseases including glaucoma, diabetic retinopathy etc and even predicting system diseases including heart failure, stroke and Parkinson’s.
‘Oculomics’ is a new field of science to identify general health disorders through the eyes. The term was first coned by Professor Alastair Denniston in 2020, who was one of co-authors of the paper.
The ‘eye’ is often called a ‘window’ that gives sneak-peaks into our general health, offering a non-invasive look at our nervous system. Having in-depth knowledge about the eye-body relationship is a key to diagnosing and treating complex diseases and other problems related to ageing.
RETFound has proved to be equally effective in successful diagnosis of different diseases in diverse population.
Lead author of the study, Ph.D. student Yukun Zhou (UCL Center for Medical Image Computing, UCL Medical Physics & Biomedical Engineering, and Moorfields Eye Hospital), says, “By training RETFound with datasets representing the ethnical diversity of London, we have developed a valuable base for researchers worldwide to build their systems in health care applications such as ocular disease diagnosis and systemic disease prediction.”
RETFound was trained on a curated database of 1.6 million images sourced from Moorfields Eye Hospital. This employed AI tools and infrastructure provided by INSIGHT, the world’s largest bioresource of ophthalmic data and the NHS-led health data research hub for eye health based at Moorfields.
The hub’s AI capabilities and powerful computing become more advanced from a 2016 research partnership between Moorfields and DeepMind, now Google DeepMind.
Yukun Zhou and Professor Pearse Keane of Moorfields and UCL, led the research team that has now made their model available on GitHub for free. Researchers worldwide, especially in China and Singapore, have been employing RETFound for their novel investigation into eye diseases.
