Using artificial intelligence for early detection of keratoconus
CERA researchers are investigating the use of artificial intelligence to help detect signs of keratoconus, thanks to new funding from the Perpetual 2020 IMPACT Philanthropy Application Program.
CERA Senior Research Fellow Dr Srujana Sahebjada is devoted to improving the quality of life for people with keratoconus, a condition that affects the cornea, the clear front window of the eye.
Keratoconus usually affects teenagers and young adults. For people with this condition, the cornea gets thinner over time and develops a bulging cone-like shape, which causes vision problems. In advanced cases, a corneal transplant will be required to correct or restore vision.
One of the challenges of keratoconus is that it is difficult to diagnose in the early stages. With new funding from the Perpetual 2020 IMPACT Philanthropy Application Program, Dr Sahebjada is leading research into a solution that could help change this, using artificial intelligence and advanced imaging technology.
“The goal of this project is to accurately diagnose and detect keratoconus patients at the earliest possible time,” says Dr Sahebjada.
“This would allow patients the best opportunity for early treatment and reduce the need for corneal transplantation.”
When keratoconus is detected early, treatments are available that can help slow the progression of the disease. This now includes collagen crosslinking (CXL), which stiffens the cornea to stop it losing its shape, preventing future vision loss.
“Patients with keratoconus are young – as young as four years – and can have poor quality of life. They represent a greater economic and societal challenge than many other eye diseases,” Dr Sahebjada explains.
“A recent study from the Australian Corneal Graft Registry estimated that keratoconus patients typically require up to five corneal transplants during their life.”
Advanced technology for early diagnosis
Keratoconus is diagnosed using topographic images – special images that show the detailed shape of the cornea, similar to topographic maps of mountains.
In a routine clinical examination, diagnosis relies on the eye specialist subjectively identifying subtle changes in these images. This means early-stage keratoconus can be easily missed.
For this project, Dr Sahebjada is combining artificial intelligence with advanced imaging technology, using a large collection of advanced topographic images of patients’ corneas that was collected over the past eight years as part of the Australian Study of Keratoconus.
“This will result in an objective screening tool that can detect keratoconus early, without requiring expertise in interpreting corneal images,” Dr Sahebjada says.
Improving access to keratoconus care
Dr Sahebjada hopes the outcome of this research could help improve keratoconus diagnosis and management in communities where access to specialist eye care is limited.
“The AI algorithms developed in this project can be incorporated into any corneal imaging system,” she says.
“This could vastly improve the diagnosis of keratoconus globally, particularly in rural clinics that would otherwise have no access to expensive advanced imaging systems.”
The impact of philanthropy
Dr Sahebjada says philanthropic support is critical for researchers studying diseases like keratoconus which are not as well known in the community.
“I am incredibly grateful for this funding from the Perpetual 2020 IMPACT Philanthropy Application Program,” Dr Sahebjada says.
“Our goal is to stop keratoconus progression at an earlier stage of the disease process, maintaining each patient’s vision and minimising the effect of keratoconus on their quality of life. This generous support will help us achieve this.”