In a world increasingly devoted to the unflagging demands of the algorithm, ophthalmology can occasionally seem a little behind in its uptake of artificial intelligence (AI) and its associated tech. This is particularly true when it comes to applying AI and deep learning models to advancing patient care in real-world clinical use.
To bring ophthalmology up to speed, as well as position the UK as a world leader in AI-enabled healthcare, Pearse Keane – consultant ophthalmologist at Moorfields Eye Hospital and Professor of Artificial Medical Intelligence at University College London (UCL) Institute of Ophthalmology – co-founded Cascader Ltd in April 2025.
Here, he tells The Ophthalmologist what he hopes this new medical technology company will achieve.
What is Cascader’s mission?
Cascader is dedicated to developing and applying AI in healthcare, with a particular emphasis on ophthalmology. The company is a start-up founded by myself, Peter Thomas (a fellow consultant at Moorfields), and Ege Ilicak, who has over 15 years of experience in medical engineering, particularly in setting up quality management systems and getting regulatory approvals. For Cascader, that's absolutely key because one of the biggest challenges is how to take a piece of experimental code published in a research paper, and then reconstruct it within a quality management system so you can meet good industry practice and all the regulatory and technical requirements.
The company is a strategic partnership between Moorfields Eye Hospital, UCL Institute of Ophthalmology (which has huge expertise in every aspect of ophthalmology and vision, science research and in particular AI), and Topcon Healthcare, one of the leading producers of ophthalmic imaging systems.
What was your initial motivation in setting up Cascader?
I think of developing medical AI systems in two categories: one is going from an idea to an algorithm, the other is going from code to clinic. Most of my career has been focused on going from an idea to an algorithm – getting an idea, compiling the data, working with good computer scientists and engineers and clinicians to develop a research algorithm, and then trying to publish in a respected journal.
So far I haven't been involved in anything that is actually incorporated into real-world clinical use, precisely because of the challenges involved in going from code to clinic. With Cascader, I was motivated by the idea of working with a great team, coming up with an idea, developing a medical AI model that is really impactful, publishing it in a journal like Nature, and then having it used by millions of people all around the world to improve healthcare. That's my career goal, and I think this company is going to be central to making that happen.
From a practical perspective, what would this translation of cutting-edge AI research into everyday clinical care look like?
The most important aspect of Topcon’s involvement is that they can bring expertise in the development and translation of medical algorithms. They have a huge amount of technical and regulatory expertise, and they are a global company that already deploys hardware and software in this space.
Ultimately, to translate these AI systems to the real world you need to have a close partnership with an imaging company, because you need to think about how these systems can be implemented. For example, are they going to be deployed locally or in the cloud?
Integration is also key, because for these systems to be successfully adopted they have to be really seamless for ophthalmologists and optometrists to use. If an ophthalmologist has to wait five or 10 minutes for an algorithm to give an answer, then that's going to be a hard sell in most cases.
Do you already have specific AI-based devices currently in the pipeline?
We would like to try and build on the work that we've previously done with Google DeepMind. In 2018 we published a study in Nature Medicine to examine how we could develop an AI-based system for triaging macular disease.
This would be a system to assess for macular and retinal disease; it could be a transformative tool for optometrists in the community, general ophthalmologists, or even specialist ophthalmologists in other areas like glaucoma.
As an example, in a UK setting, imagine you have difficulty reading, or you have some distortion in your vision. You go to a high street optometrist for an eye test, and they will often offer an OCT scan as part of the test. My hope is that this particular AI system could be deployed in the community and identify patients with sight-threatening macular disease at the earliest point. Those patients could then be prioritized and referred to a retinal specialist to get any treatment that they need.
Is Cascader primarily UK-focused for the moment?
We're very keen to try and deliver something that can benefit patients in the UK, but of course we're also interested in global patient benefit. I don't think that those things are mutually exclusive. Once we've figured out the UK, it would be wonderful to go global and impact as many people around the world as possible.
How does oculomics (i.e., the use of advanced retinal imaging to uncover insights into systemic conditions) factor into Cascader?
We've known for more than 100 years that we can use the eye as a window to the rest of the body. But this idea has been supercharged in the last few years, with big data, advanced imaging techniques like OCT, and the latest advances in AI.
We are in a very powerful position to explore this because at Moorfields and UCL we have linked large-scale eye imaging data to an NHS database of hospital episode statistics. In the medium-term, we are interested in moving beyond macular and retinal disease to look at other conditions, like glaucoma and neurodegenerative diseases such as Alzheimer's and Parkinson's. The ultimate goal would be to have an AI system that could assess a routinely collected retinal image to predict a person’s risk of heart attack or stroke, or to make an early diagnosis of metabolic diseases and other systemic conditions.
If we could pick up systemic disease at an earlier stage and get people to see their family doctor and have all the other appropriate tests, that could be a world-changing thing.
Do you feel like current technology is close to being able to identify systemic diseases?
I think it's close, but for me it's still in the research stages and we're keen to translate it. There are a number of companies who have raced ahead and are already exploring regulatory approvals. We're being a little more cautious. We want to make sure that we have algorithms that perform robustly and are highly accurate, but also that we've thought through all of the ethical considerations. For example, is it appropriate to tell someone that they might have early signs of Parkinson's or Alzheimer's? And if so, what are the next steps for that person? We need to carefully examine these sorts of questions.
Is there anything else you would like to add?
I'm Irish and I trained in the US, but the reason I've come to the UK is because I passionately believe that the UK can be a world leader in AI-enabled healthcare, not just in ophthalmology but more broadly speaking. The research work we've done so far is a good example of how you can combine world-leading data from Moorfields and world-leading computer science from UCL to advance medical research. For me, the formation of Cascader is about the next stage, and how we can turn that research into world-leading innovation. It's one thing to publish good research papers; it's quite another thing to deploy these systems at scale. I am determined that this company will be an exemplar of how that might be achieved.
