Artificial intelligence to diagnose causes of chronic optic neuropathy
Brief plain language background
Our eyes are essential to vision; but our brain must process and interpret the visual information collected by our eyes before we ‘see’ an image. This visual information is transmitted from the back of our eyes to the brain by the optic nerve; and so, optic nerve damage can cause permanent vision loss.
Optic nerve damage is common; indeed, optic nerve damage caused by high pressure inside the eye, known as glaucoma, is the most common cause of irreversible sight loss globally. But there are many other disorders, besides glaucoma, that can damage the optic nerves, including but not limited to: inadequate blood flow, inflammation, infection and tumours.
Doctors can struggle to tell different causes of optic nerve damage apart because they look the same on eye scans and vision tests. This can lead to delayed or missed diagnosis of life-threatening causes, like brain tumours.
What problem/knowledge gap does it help address
Traditionally, doctors make a diagnosis based on how symptoms developed over time (as told by the patient), the appearance of the optic nerves, vision and blood test results, eye and brain scans, and the patient’s medical history. This process can be slow because there is often a delay before scan reports and test results become available. Additionally, there are relatively few doctors specialising in the diagnosis of optic nerve problems and their service is not offered by every hospital in the country, which can mean some patients get the wrong diagnosis, leading to delays in treatment that could save their sight or lives.
Artificial intelligence (AI) systems can be trained to recognise patterns in complex data, like eye and brain scans, helping doctors diagnose optic nerve disorders. So far, no-one has trained AI systems on a large enough set of data covering all the different causes of optic nerve damage.
Aim of the project
To train an AI system to recognise and label different causes of optic nerve damage from it.
Potential impact on people with sight loss
Within 2–3 years, the AI-assisted analysis of eye scans and vision test results will help doctors better understand how different disorders lead to different patterns of optic nerve damage. This knowledge can guide doctors when trying to diagnose patients with optic nerve damage.
In 5–10 years, AI could be used in optician practices and hospital eye clinics to help eye professionals spot problems earlier and more accurately. This means faster diagnoses, fewer unnecessary tests or appointments, and earlier treatment where needed. For patients, this brings peace of mind and the best possible chance of protecting their vision.