Walter and Eliza Hall Institute of Medical Research in Victoria, Australia led one of the largest eye studies in the world that used AI to analyse eye images of over 50,000 people to better understand the retina's connection with various diseases.
The international study, funded by California-based Lowy Medical Research Institute, tapped into a dataset of optical coherence tomography (OCT) images from around 54,000 individuals stored at the UK Biobank. Utilising a convolutional neural network, it generated "the highest-resolution spatial dataset of retinal thickness ever produced," creating 50,000 maps with measurements at over 29,000 locations across the retina.
Moorfields Eye Hospital and University College London from the United Kingdom and the University of Washington in the United States were collaborators in the study.
FINDINGS
While previous studies have indicated correlations between retinal thickness and diseases, the WEHI-led research provided a deeper look into the complex anatomy of the retina.
One of the significant findings of this study, published in Nature Communications, is that reduced retinal thickness is "highly associated" with multiple sclerosis (MS), a chronic neurological disease affecting the brain and spine. "This result provides strong, independent confirmation of multiple reports of the utility of OCT as the source of biomarkers for MS and MS progression," it added.
Retinal thinning is also highly associated with a range of other neurodegenerative diseases, as well as cardio-metabolic disorders.
"We illustrated that the retina has unique metabolic sensitivities, with retinal thickness associated with multiple systemic metabolic diseases, and metabolites previously implicated in several retinal diseases," the study explained.
Additionally, the study identified genetic factors that influence retinal thickness, suggesting that at least 294 genes play a role in the retina's growth and development.
WHY IT MATTERS
Researchers claim their study opens up the potential for using routine eye imaging to screen for and manage diseases. Neurodegenerative conditions like dementia and metabolic disorders such as diabetes are linked to degeneration or disruption of the central nervous system, which the retina is a part of.
"We’ve shown that retinal imaging can act as a window to the brain, by detecting associations with neurological disorders like multiple sclerosis and many other conditions," lead researcher Dr Vicki Jackson from WEHI.
The study specifically points to the potential for retinal thickness as a diagnostic biomarker "to aid in detecting and tracking the progression of numerous diseases."
"We can now pinpoint specific locations of the retina which show key changes in some diseases," Dr Jackson said.
Moreover, the study contributes to the growing field of oculomics as a "non-invasive approach for predicting and diagnosing diseases."
THE LARGER TREND
Another Australian study has linked the damage to the conjunctiva of the eye to pterygium, or tissue growth on the cornea, which can be an early predictor of skin cancer. A desktop and mobile AI-driven detection system has been developed to assess sun-related eye damage.
The Singapore Eye Research Institute has also utilised AI to scan people's retinal photos and assess their health conditions. Over the past five years, it developed two novel solutions: one for screening chronic kidney disease and another for predicting a person's biological age.
Meanwhile, in China, a generative AI model was built for automated eye disease diagnosis. Called VisionFM, it was pre-trained using 3.4 million eye images from over 500,000 people worldwide. Besides diagnosing eye diseases, it can be applied to eye disease progression prediction, systemic biomarker prediction through ocular imaging, intracranial tumour prediction, and lesion, vessel, and layer segmentation.