Google’s DeepMind is training AI to diagnose eye disease

Findings of a collaborative project between DeepMind and London’s Moorfields Eye Hospital have shown “promising signs”

Artificial intelligence could be used in hospitals to provide early diagnosis of eye disease after collaborative research between Google’s DeepMind and a world-leading eye hospital showed “promising signs”.

DeepMind, the London-based AI company owned by Google, teamed up with Moorfields Eye Hospital and analysed thousands of retinal scans to train an AI algorithm to find signs of eye disease.

Moorfields said the findings had been sent to a medical journal, which they hoped would be published by the end of 2018. If the results pass an academic peer review, clinical trials could start within the next few years.

“Research shows that an ageing and growing world population could lead to a threefold increase in blindness by 2050, so it is vital we explore the use of cutting-edge technology,” said Professor Sir Peng Tee Khaw, director of research and development at Moorfields.


Read more:


“If we can diagnose and treat eye conditions early, it will give us the best chance of saving people’s sight. Machine learning could support healthcare professionals to quickly analyse highly complex optical coherence tomography (OCT) scans to diagnose retinal diseases. This means that patients with sight-threatening conditions could be prioritised and seen urgently in hospital eye services.”

At the moment, retinal scans are examined by specialists, which takes a considerable amount of time.

The AI algorithm is being trained to spot signs of glaucoma, age-related macular degeneration and diabetic retinopathy, which are three of the biggest eye diseases in the world, quicker and more efficiently than a human specialist.

Prof Tee Khaw said he was "optimistic" that the research could prevent people around the world from suffering avoidable sight loss.

DeepMind has also partnered with University College London Hospitals and Imperial College London to develop its algorithm for eye scans for radiotherapy scans and mammograms, the Financial Times reported.

“In areas like medical imaging, you can see we’re going to make really tremendous progress in the next couple of years with artificial intelligence,” DeepMind Health’s clinical lead Dominic King told the paper.