Here we look at artificial intelligence and opthalmology, with an introduction to AI for eye health including its existing uses in identifying medical issues and potential improvements in future care and treatments.
Artificial intelligence is often defined as the theory and development of a computer system able to perform tasks normally requiring human intelligence. Tasks such as visual perception, speech recognition and decision-making.
Although AI seems as it has been discovered only yesterday and the vast unknown of AI has yet to be discovered, AI has in fact been around since the mid-20th century. British logician and computer pioneer Alan Mathison Turning described an abstract computing machine, this was the earliest theoretical work.
The earliest successful AI programme was written in 1951 by Christopher Strachey. Eventually, at a conference at Darthmouth College, Hanover, New Hampshire, the term “artificial intelligence” was coined.
AI already plays an important role in our daily lives and impacts our routine and takes even if we are not aware. Examples of AI in day-to-day lives include opening a phone with face ID, e-mails and social media and even Netflix recommendations. Likewise, big corporations and industries such as banks and Siemens also use AI. Services such mobile banking and cloud – based system are all based on AI and machine learning. According to a PwC study, in 2018 roughly 62% of large companies were already taking advantage of AI technology.1
Existing Use of AI in Medicine
In the field of medicine – AI is already a big part of the medical landscape, ranging from scheduling patient appointment online to digitalisation of medical record and reminder calls for follow – up appointments. In a more advanced form of AI being used, some studies have been documented where AI systems were able to outperform dermatologists in correctly classifying suspicious skin lesions.
This is due to the AI system being able to learn more from many different cases and is exposed to multiple cases within minutes. Although AI has many advantages including efficiency, accuracy and precision, a rather disheartening disadvantage is the potential loss of jobs. However, AI has a big part to play in the medical field in the future and the aim is to find a balance between the effective use of automation and AI and the human intuition and judgment.2
Artificial Intelligence in Eye Health Diagnostics
For Ophthalmology and Optometry, AI is being used in hospital settings, with the aim that in primary care settings such as independents and multiples will soon have access to AI in their location. Currently, many Optometrists have access to Optical coherence tomography scan.
OCT is an imagine technique which uses low-coherence light to capture 2D / 3D images from biological tissue. It allows Optometrists to view the eye in greater detail and see what is happening below the surface. OCT scans help detect sight-threatening eye condition much earlier, one claim is that with OCT scans, glaucoma can be detected 4 years earlier.
Although OCT scans show what the eye in more details, interpreting is quite difficult and takes time to fully understand and potentially diagnose.
In hospitals, OCT scans are also performed. Moorfields Eye Hospital an estimated 5,000 OCT scans are performed, with such a large volume of scans to analyse and take action depending on the conclusion, this can impact on how quickly patients can be seen to discuss their diagnosis and treatment. In 2016, Moorfields Eye Hospital NHS foundation trust entered into a research partnership with the British company DeepMind.
DeepMind was later acquired by Google Health and Moorfields decided that they would continue the partnership with Google Health. One aspect of their partnership is to see how AI technology could help analyse the OCT scans.3
In 2018, a research article was published in Nature where researchers had created a novel AI programme / technology that could interpret the results of OCT scans to the level of expert clinician, was published. Researchers from Moorfields and associate groups developed the artificial intelligence technology, which has the accuracy level of specialist ophthalmologist and optometrists when identifying a wide range of eye related conditions.
Similarly, the research conducted found that the AI technology has an opportunity to be used in real-life healthcare settings. Similarly, the technology can also provide information that explains how it has come to the decision it has, this Moorfields says it critical as it will allow healthcare professionals to look over the recommendation in detail.
Lastly, Moorfields found that the AI programme is compatible with a number of scanning machines and this able to analyse more scans even in primary care settings. This increases by tenfold the number of people it could benefit, whilst the AI programme becomes future proof against newer devices and models.4
The system works using two types of neural network. Mathematical systems for identifying pattern in images or data, the AI system quickly learnt to identify ten features of eye disease from highly complex OCT scans 5.
Reena Chopra, an Optometrist who was part of the Moorfields and DeepMind project, said that the researchers had to spend many many hours outlining the different features in hundreds of OCT scans into the AI programme, spending a lot of time making sure that they were pixel perfect. Ms Chopra described the importance of having high-quality data when it comes to AI, putting it simply as a baby learning new things, it will only learn from the data it is fed by you. If incorrect data is given, then it will not learn anything.6
The study concluded that the AI programme was able to make the correct referral recommendation more than 94% of the time for over 50 eye diseases, this matched the performance and recommendation of expert clinicians.
Although it is still early and the research is still going on, there is strong belief that AI could transform the way several eye diseases are diagnosed, treated and managed. Thus, a system where patients with the most serious eye diseases are prioritised before irreversible damage is caused. As such Dr Pearse Keane, Ophthalmologist, said that AI has the potential to completely change the way diabetic retinopathy screening is performed.7
AI Trial at Moorfields Eye Hospital with Dem Dx
Likewise in January 2021, Moorfields Eye Hospital partnered with healthcare company Dem Dx, to develop an AI – powered system. This new system will look to help and support paediatric nursing staff as they embark on taking greater responsibility for initial triage and diagnosis.
The trial of this new AI-powered system will see 13 paediatric nurses at Moorfields trained at using this new system, which will help the nurses triage conditions such as various types of conjunctivitis, cellulitis and newly found squints. The support and training from Dem Dx will aim to assist with clinical decision making, allow nursing team to take on additional clinical responsibilities with more efficiency and safety. It is hoped that these changes will lead to earlier and more accurate detection of common eye diseases in children, less waiting times for patients.8
AI does seem to hold a lot of promise and is highly thought of, however concerns about what will happen to the human workforce is an important question. According to one AI technology company’s CEO, optometrist should have nothing to fear, further expressing that even with AI playing a role in the future, the patient will still desire the human touch. The patient will still yearn the human contact.
This is evident with the British public’s shopping habits, even though online shopping is fast growing and present, the Brick and mortar still continue to be visited by thousand of customers. Likewise, the same CEO also suggested that AI can empower Pptometrists, giving them the tools Ophthalmologist have, thus improving their accuracy with diagnoses and making better decisions regarding referrals. 9
Potential of AI for Patient Care
Ultimately, AI has a vast potential to help patients far beyond the scope of Ophthalmologists and Optometrists. With quick learning and immense amount of storage and memorising capabilities, AI technology and machine learning offers a great number of solutions for patients. In a recent discussion with The College of Optometrists, Ms Chopra discusses another project where her and a group of researchers trained an AI-system to predict whether a patient with wet AMD in open eye will develop the condition in their second eye 10.
This sort of prediction and analyses is ground-breaking and can have an ever-lasting impact on the patient. For now, all AI system will need to go through rigorous clinical trials and regulations but as Dr Pearse Keane said, automation and this sort of technology is certainly close and not too far away.
Subsequently, this sort of technology has the prospect of impacting many people around the world and if utilised correctly and with the empathy, experience and direction of a practitioner, AI can make an ever-lasting impact.
Written by Mohammed Taherali, who is a graduate in Opthalmology from the University of Plymouth, UK.