Harnessing the power of AI for eye health
At Moorfields Eye Charity, we are proud to be funding world-leading Artificial intelligence (AI) innovation in vision research and eye health.
Artificial intelligence, or AI, is an advancement in computing where computer systems learn from data, identify patterns and perform tasks that we would usually consider as needing human intelligence, such as pattern recognition informed decision-making.
In doing so, AI can speed up many of the processes that take time for humans to complete. This has crucial applications in ophthalmology, including:
- untangling fundamental features of eye biology and disease
- assisting diagnoses
- informing treatment options
Therefore, AI is an important tool in the modern research arsenal that has the potential to revolutionise our knowledge of the eye and transform healthcare.
Together, Moorfields and the UCL Institute of Ophthalmology have identified three key areas for driving advancements in ophthalmology. These are interlinked research disciplines that will play a vital role in transforming scientific knowledge and clinical care delivery across different disease areas and experimental research programmes.
Excitingly, datasets from all these areas have the potential to be utilised and rapidly advanced by AI approaches. The three priority areas are:
Genomic medicine is an emerging medical discipline that involves analysing an individual’s DNA to inform their clinical care. Genomic data helps scientists gain new insights and understanding about the causes of disease and find better ways of helping patients, especially those who have conditions that are currently hard to treat.
Imaging in ophthalmology allows visualisation and understanding of the eye and the body in greater detail than ever before. With rapidly advancing technology and new techniques we are gaining incredible insights into how the eye works. Imaging is particularly powerful in ophthalmology due to the external accessibility of the eye, which means that a lot of information can be gained through non-invasive imaging techniques.
Health informatics is the intelligent use of information and technology to provide better care. Excellent patient care depends on the fast and accurate flow of information. With a vast patient base at Moorfields and outstanding research talent, we have an opportunity to build on our highly impactful research with large scale data and advanced analytics to enhance the diagnosis, prevention and treatment of disease.
Here are some of the exciting projects that Moorfields Eye Charity are proud to be supporting in the field of artificial intelligence (AI).
Moorfields Eye Charity investments in AI projects
AI has the power to rapidly inform and innovate clinical practise. In one clinically-focussed project, associate professor Anthony Khawaja is uniting the fields of genetics and informatics to develop effective risk stratification and predictive tools in glaucoma.
Glaucoma treatments could then be tailored to each patient to preserve sight in high-risk individuals, whilst reducing treatment-associated side effects and costs for lower risk patients.
We are excited to also be supporting Professor Pearse Keane’s pioneering use of AI. Most recently, we have funded his work with honorary clinical research fellow Dr Edward Korot on using AI deep learning to classify clinical images of the eyes of patients with diabetic retinopathy, age-related macular degeneration (AMD) and other common retinal diseases.
Dr Korot kindly agreed to talk to us about his research and how AI has the potential to improve the lives of patients with eye conditions.
Findings from this research have already led to three recent publications in the journals of Lancet Digital Health, Nature Machine Intelligence, and Scientific Reports.
Dr Korot’s recent Nature Medicine Intelligence publication highlights the potential of automated AI in the clinic. This would reduce the burden on clinicians to understand computer programming. However, given the implications of using AI in delivering critical clinical care, the researchers also stress that the technology should be able to report its methods to the user to enable double checking of important conclusions.
Three different approaches to studying inherited retinal diseases (IRDs)
Thousands of clinical eye scans are undertaken each week at Moorfields alone, which provide a huge amount of data that can feed into many AI projects. Moorfields is also very fortunate to have a wealth of electrophysiology data.
Visual electrophysiology is a collection of advanced techniques used to test the functioning of cells along the full visual pathway – from the retina to the optic nerve to the primary visual cortex in the brain.
In their electrophysiology-based project, Dr Anthony Robson and Dr Omar Mahroo are applying AI computational learning to better understand Stargardt disease.
Stargardt disease is a rare condition caused by degeneration of the macula – the central area of the light-sensitive retina at the back of the eye.
Electroretinograms (ERGs) are commonly used to assess Stargardt disease by measuring the electrical activity of the retina. The team will investigate whether AI can be used to reveal subtle information and ERG changes that might not otherwise be spotted.
Additionally, we are supporting the work of Dr Adam Dubis and Dr Watjana Lilaonitkul, from the UCL Institute of Ophthalmology and UCL Institute of Health Informatics, respectively, on rare inherited retinal diseases (IRDs).
Dr Dubis and Dr Lilaonitkul have pulled together an international inherited eye disease dataset, which AI techniques can then use to predict how the structure of the retina changes over time in individual patients with IRDs.
1 in 2,000
Estimated incidence of inherited retinal diseases, the leading cause of vision loss in patients between ages 15-45 years
The ability to predict individualised outcomes of treatment and disease over time can potentially enable future clinical trials to become cheaper, faster, and more effective. Until new treatments become available, the technology developed here could also help doctors provide better care and more detailed advice to patients with IRD.
AI technologies like this can help revolutionise the landscape of rare disease studies, shorten the timescales for developing novel therapeutics and offer new hope for patients.
This project brings together amazing international collaboration and leverages the concepts of team science to improve patient care and potentially help save sight.
Dr Adam Dubis, UCL Institute of Ophthalmology (below left)
At the heart of our research lies the hope to better the odds for patients with IRD. We are very grateful for MEC’s support towards our commitment to innovate new technologies for this cause.
Dr Watjana Lilaonitkul, UCL Institute of Health Informatics (above right)
Dr Nikolas Pontikos, a Research Fellow and group leader at the UCL Institute of Ophthalmology, has pioneered using AI to tackle genetically diagnosed IRDs. His Career Development Award (CDA) from MEC has focussed on developing an innovative new tool called Eye2Gene.
Eye2Gene is trained with retinal scans and genetic data from Moorfields and UCL, using the largest and most complete IRD dataset in the world, as published in Dr Pontikos’ paper in the journal Ophthalmology.
The end goal for Eye2Gene is to be able to automatically suggest the gene and mutations that are causing an IRD from a retinal scan. Using technology such as Eye2Gene therefore aims to assist genetic testing by making the genetic diagnosis of IRDs more widely available and more efficient.
We are also delighted to announce that, thanks to the support of a CDA from us, Dr Pontikos has recently been awarded a £1.3 million grant from NIHR AI Award to develop Eye2Gene into a medical device within the NHS.
What’s next for informatics and AI research?
Researchers and clinicians at Moorfields and UCL are dedicated to using informatics to drive innovation and breakthroughs at the forefront of our understanding and treatment of eye conditions.
The projects described above highlight how informatics, imaging, and genomic medicine interweave to help, one day, enable greater personalised medicine options, increase the accuracy of diagnoses, and support the development of more treatments options.