At this year’s Vis-Ability (Fight for Sight’s annual conference) in Westminster, leading voices from research, healthcare and politics came together to ask an important question: how can innovation in eye health improve outcomes for everyone – not just those who are easiest to reach?
With advances in technology a simple eye test can now tell us far more than our latest prescription. We know from research we have funded, that changes occur in the eye of patients’ years before their brain shows any sign of change.
Retinal imaging and AI are making it possible to analyse millions of images and spot differences in these images that are too subtle for humans to see, at a scale that would take humans years to achieve.
During the panel discussion From Sight to Insight: Innovating Eye Health for Everyone, Josef Huemer, a consultant ophthalmologist and research fellow explained why he believes the potential of retinal imaging goes well beyond diagnosing eye disease.
That ability to work with very large numbers of retinal images could open up new opportunities for researchers to understand the relationship between what we can see in the eye and what is happening elsewhere in the body.
“We have the ability to connect imaging – retinal imaging – at scale.”
What can the retina tell us?
The retina is unusual because it provides a direct view of blood vessels and neural tissue. This makes it potentially valuable not just for understanding eye disease, but for studying wider health.
As Josef explained, researchers are already linking retinal images with other sources of health information to investigate these relationships.
This kind of linked data allows researchers to ask questions that would be difficult to answer using eye images alone.
Could changes in the retina be associated with neurological disease? Could patterns in the eye help us understand what is happening in the brain? Could retinal imaging eventually contribute to earlier identification of conditions that currently require very different tests to diagnose? The potential is significant.
Josef described the opportunity to analyse the retina in much greater detail, looking at different parts of the retina and investigating how they vary across diseases.
AI could help us see patterns we cannot see ourselves
The scale of the data is important. A clinician can examine an individual retinal image and identify many important features. But AI can analyse enormous numbers of images and look for patterns across them.
Josef pointed to evidence that retinal photographs already contain more information than we might expect.
“We already know that from a photograph the back of the eye (using OCT scans) we can predict so much from the biological data, to whether someone is a smoker or non-smoker.”
The next challenge is understanding what those signals mean, whether they are reliable and whether they can be translated into useful healthcare applications.
That distinction matters. Being able to identify a statistical association in an image is not the same as having a clinically useful diagnostic test.
The research needs to establish whether these technologies can make a meaningful difference to patients – and whether they can do so safely and equitably.
From Fight for Sight funded eye research to wider health
Fight for Sight has already supported research exploring this potential.
Professor Pearse Keane and colleagues created AlzEye, a database linking millions of retinal images with healthcare data from more than 150,000 people. The project was supported by a £15,000 grant jointly funded by Fight for Sight and Alzheimer’s Research UK. Researchers have subsequently used the database to investigate links between retinal changes and conditions including Parkinson’s disease, heart disease, stroke and dementia.
One striking finding from this work was that changes in the retina associated with Parkinson’s disease could appear up to seven years before other symptoms. The researchers are continuing to investigate what retinal changes might reveal about other diseases.
Fight for Sight has also supported the SCONe Project, which has created a repository containing one million retinal scans from community optometry practices. The project is designed to support research using retinal images and AI, with the aim of improving understanding, diagnosis and monitoring across healthcare.
These projects show why investing in the underlying research infrastructure matters. Progress does not necessarily begin with a finished diagnostic tool. It can begin with building the datasets, developing the technology and asking the right questions.
The potential is not limited to common conditions
Dr Shokat Adam MP highlighted another important opportunity: using AI and retinal imaging to improve our understanding of eye conditions (link) that are currently difficult to diagnose or treat.
He pointed to diabetic retinopathy (link) as an example of what can happen when technology is applied at scale.
“There are automated systems that can detect, grade and diagnose altogether,” he said.
The experience of diabetic eye screening could potentially provide lessons for other areas of ophthalmology.
Shokat also highlighted the possibility of using AI to identify eye conditions earlier, including rare genetic disorders.
“There’s exciting potential for these aspects including rare genetic disorders,” he said. “Something that is difficult to get funding for because it is rare singularly, but collectively they are not rare at all and cause a huge impact.”
This is an important point for research more broadly. Technologies and approaches developed while studying one condition can create tools, knowledge and datasets that benefit work in other areas.
Fight for Sight is currently funding research using AI to investigate conditions including Birdshot chorioretinopathy and chronic optic neuropathy. In the Birdshot chorioretinopathy project, researchers are developing an AI screening tool that could help identify the rare eye condition from retinal photographs and combine this with genetic and clinical information.
Meanwhile, research into chronic optic neuropathy is exploring whether AI can help distinguish between different causes of optic nerve damage using eye scans and other clinical information.
Discover more about eye conditions including inherited retinal disease in our A to Z.
From possibility to patient benefit
The excitement around this research needs to be matched by careful testing.
Josef acknowledged that researchers need to understand how technologies perform across different populations, while Shokat stressed the importance of ensuring that AI systems are trained on representative data.
The challenge is therefore not simply to develop AI that can analyse an eye.
It is to develop technology that works for the people who need it, and to understand how it can be safely integrated into healthcare.
That is why events such as Fight for Sight’s Vis-Ability matter. Bringing researchers, clinicians, policymakers and people with lived experience together creates an opportunity to consider not just what technology can do, but what it should do and how it can benefit patients.
The next step: investing in what comes next
The idea of the eye as a window into wider health is no longer simply a fascinating scientific theory. Research supported by Fight for Sight and others is helping to establish how retinal imaging, large datasets and AI could contribute to earlier diagnosis and a better understanding of disease.
But there is still a long way to go.
The technologies need further research, validation and testing. Researchers need access to representative data. And new discoveries need investment if they are going to move from promising research into tools that can make a difference to patients.
That is where support for research matters.
"I can't thank the eye researchers enough. I mean, it's the reason why I've got a positive outset on life now."
Tommy Salisbury
Related content
