We often think of modern medicine as being cutting edge and using sophisticated equipment.
But when looking at the back of the eye, doctors often use a tool more than 170 years old.
Dr Hong Sheng Chiong, co-founder of social enterprise oDocs Eye Care, said looking at the retina through the pupil was done using an ophthalmoscope, created by German physicist and physician Hermann von Helmholtz in the 1850s.
"He created this small instrument using a candle as the source of light with a mirror to reflect the light into the person’s eye, and with a small aperture you can then see through."
This was like trying to see a room by looking through a keyhole, Dr Hong said.
"It struck me, why haven’t we changed the way we look at the back of the eye?"
In 2014 Dr Hong and his team began to develop a solution — an ophthalmoscope that could connect to a smartphone.
"So instead of looking into a room through a tiny keyhole, it is like looking into a room through a big window."
Once it was possible to take a picture of the retina, it could be sent to colleagues and other specialists to help diagnose symptoms.
"That is what we are trying to pioneer, teleophthalmology, or telemedicine in ophthalmology."
This could generate lots of pictures and data on the retina, so Dr Hong thought there might be a more efficient way of analysing them.
If a clinic was doing diabetic retinopathy screening there would be thousands and thousands of photos being generated each month.
His team started thinking about using artificial intelligence (AI) to analyse the data automatically.
Using an open-source network called Inception, Dr Hong and his team created a free platform called medicmind.
"It is not a single product, but it is actually a platform that allows clinicians and researchers around the world to create and train their own AI."
Some projects had included training artificial intelligence to recognise diabetic retinopathy and glaucoma, he said.
Medicmind enabled GPs anywhere in the world to connect to an ophthalmologist to help diagnose a patient’s needs.
It also could be used for other data sets that were based on a collection of images.
Basically, the platform provided the tools, ease of viewing and ability to develop AI for any field, not just the eyes, Dr Hong said. Any image based on a data set could be used, for example a set of paediatric X-rays.
"You get 1000 kids with pneumonia and 1000 without and you can easily drag and drop using our platform and create an AI without knowing programming language."
