UK-based Scientists Unveil RETFound to Diagnose Health Conditions from Retinal Images

RETFound learns by analysing millions of retinal images, accurately predicting missing parts, and developing a deep understanding of a healthy retina's intricate features from diverse image exposure.

In a recent development in the field of artificial intelligence (AI), London-based scientists have unveiled AI tool- RETFound. This tool is developed to diagnose and predict a wide range of health conditions, including heart failure, through the analysis of retinal images.

The power of AI is rapidly permeating various sectors, from the corporate world to the medical arena, offering invaluable assistance to humans by simplifying tasks and providing in-depth analytical insights.

RETFound promises to enhance efficiency and reduce the costs associated with developing AI tools for medical diagnosis.

Unlike traditional AI models that require extensive manual data labelling, RETFound autonomously learns from a dataset of 1.6 million unlabeled retinal images. This autonomous learning promises to improve efficiency and significantly reduce the development costs associated with creating AI tools for medical diagnosis.

RETFound's learning process pores over millions of retinal images to accurately predict missing portions of these images. It learns what a healthy retina should look like and comprehends its intricate features through extensive exposure to diverse images.

Pearse Keane, an ophthalmologist at Moorfields Eye Hospital NHS Foundation Trust in London, said, "Over the course of millions of images, the model somehow learns what a retina looks like and what all the features of a retina are."

Retinal Images: A Window to Health

With its intricate capillary network, the retina offers a direct view of the body's smallest blood vessels. This unique characteristic enables RETFound to provide insights into an individual's cardiovascular health.

Conditions affecting the circulatory system, such as hypertension, can be directly observed through retinal images. This capability holds tremendous promise for early detection and intervention in systemic diseases, potentially saving countless lives.

Additionally, the retina shares striking similarities with the central nervous system, resembling the brain in certain aspects. This resemblance opens the door to using retinal images in assessing neural tissue. However, interpreting these scans often requires specialised expertise, making AI an invaluable asset in this context.

While its performance in detecting eye diseases such as diabetic retinopathy can be remarkable, its accuracy in predicting systemic diseases including heart attacks, heart failure, stroke, and Parkinson's disease may not be commented on now.

In another development, Scientists at UNSW Sydney, in collaboration with Boston University, unveiled an AI tool that holds promise for the early detection of Parkinson's disease, even years before the initial symptoms manifest. Named CRANK-MS, this innovative tool harnesses the power of machine learning to scrutinise combinations of metabolites within patients' blood samples.