AI in Ophthalmology Has Huge Potential
Pairing artificial intelligence (AI) with medical imaging will revolutionise healthcare. Non-surprisingly, ophthalmology holds massive potential for AI applications since imaging is core to this medical field. In fact, the first AI device approved by the US Food and Drug Administration (FDA) was for an eye disorder.
How AI can be used within ophthalmology is far-reaching. Firstly, AI can assist in diagnosing and monitoring eye-related diseases such as diabetic retinopathy (DR), retinopathy of prematurity (ROP), glaucoma, age-related macular degeneration (AMD), cataract, and retinal vein occlusion (ROV). But that’s not all. AI has great potential in ophthalmology because medical images of the eye can show signs of other health conditions. An example is the OCTAHEDRON project. Using gliff.ai, this team aims to use machine learning to detect early signs of neurodegenerative diseases identifiable within scans of the retina.
Eye Disorders are Widespread
In many countries, providing eye care to everyone who needs it can be a challenge. This is especially the case within societies with ageing populations. Age-related macular degeneration (AMD), for example, is a condition usually present among people over the age of 50. Approximately 600,000 people are affected by AMD in the UK. Besides impairing vision, this eye condition can directly affect quality of life. For the elderly, that also means reducing their independence.
Age is not the only factor influencing demand for eye care. Between 1980 and 2014, the number of adults with diabetes worldwide has increased from 108 million to 422 million. And about 25% of people with diabetes have some form of changes to the retina.This is known as diabetic retinopathy (DR). Therefore, ophthalmologists and optometrists must often screen large (and increasing) numbers of patients for vision-related complications caused by diabetes.
Unfortunately, delays in access to medical care can risk conditions deteriorating, which may lead to significant sight loss. Moreover, certain ocular diseases require early detection to ensure the best outcomes for patients (including DR and AMD).
How AI can Benefit Ophthalmology
Efficacious AI applications for eye care can offer decision-making support to ophthalmologists, optometrists and general practitioners. As well as supporting diagnosis, AI could help inform treatment plans and predict outcomes for patients with ocular diseases. AI tools could also, for example, predict the risk of ectasia after refractive surgery.
Far and away, the most obvious way that AI can benefit eye care is by integrating it into ophthalmic screening tests, to increase efficiency. It can take several days for images to be processed by ophthalmic specialists off-site. However, AI tools could reduce the time between screening and the interpretation of images of patients’ retinas. AI tools could even offer instant feedback that can be shared with patients before they leave the healthcare setting. This in turn could lead to big cost reductions for health institutions, by reducing reliance on specialist imaging readers, and the number of patient appointments.
Ophthalmic AI applications could also ensure that expertise related to less common eye conditions are made more widely available. On the other hand, practitioners could employ AI applications to help detect common conditions, thus reducing the need to refer patients to ophthalmic specialists.
Data, Data, Data
Millions of eye tests are taken every day so there is a vast amount of medical imaging data being generated. The most commonly used ocular images are fundus photos and optical coherence tomography (OCT) scans.
Using optical scattering within biological tissue, an OCT scan creates ultra-high-resolution 2D and 3D images of the retina and anterior segment. Many eye diseases and pathologies can be assessed using this technique, including glaucoma, macular degeneration and multiple sclerosis. Previously, as Intogral Ltd, gliff.ai successfully created an OCT AI device which enabled rapid 3D visualisation and detection of macular holes and macular edema. This technology is currently undergoing further clinical investigation at Newcastle University.
gliff.ai’s software enables users to embed knowledge into high-quality, auditable datasets which can be used for building world-changing and trustworthy AI in ophthalmology. Built to the highest standards for medical AI development, gliff.ai is open-source, end-to-end encrypted, and easy to use.
Regarding AI development in ophthalmology, gliff.ai helps users to:
- Curate datasets with 1000s of images – including fundus photos and OCT scans which require pixel-level annotation tools
- Label and annotate images so that clinicians can embed their specialist knowledge into datasets
- Collaborate so that data scientists, ophthalmologists and other collaborators can come together in a single platform.
You can also read our case study about how a team at Sunderland Eye Infirmary is using OCT scans from patients with diabetes and fluid leak in the centre of the retina to create a ground-truth dataset for a future AI application.