StemCell Therapy

NewsDeepMind's AI system can spot both the presence of age-related macular degeneration in eyes and predict its progression

Tuesday, 19th May 2020, 9:57 am

Google's DeepMind research lab has developed a form of artificial intelligence (AI) that is not only capable of detecting the presence of a sight-threatening eye condition but predicting its progression over six months.

The research, conducted in partnership with Moorfields Eye Hospital and Google Health, concentrated on age-related macular degeneration (AMD), the UK's most common cause of sight loss.

While around three-quarters of patients with the condition have a relatively mild, early form called "dry" AMD, a further 15 per cent will develop exudative or "wet" AMD - which can cause permanent vision loss as a result of blood vessels leaking fluid under the eye's retina.

Early detection of wet AMD allows ophthalmologists to treat the condition, lowering the likelihood of permanent loss of a patient's central vision.

Interventional and preventative tool

The researchers developed an AI system capable of predicting whether a patient will develop wet AMD within six months, which could be a useful interventional and preventative tool.

The model was trained on an anonymised dataset of 2,795 patients who had been diagnosed with wet AMD in one of their eyes and were being regularly assessed using 3D optical coherence tomography (OCT) images of the retina.

It used two neural networks informed by raw 3D scan of a patient's eye and different tissues within the retina to develop a prediction system capable of estimating a patient's likelihood of developing wet AMD in the next six months.

When testing against the assessment skills of experts three retinal specialists and three optometrists, the system performed as well as, and in some cases better than, the clinicians in predicting the progression of wet AMD.

The findings, published in medical journal Nature Medicine, could inform doctors in developing and studying preventative treatments for eye diseases.

More data required

While the system shows promise, it would be subject to robust clinical trials and regulatory approvals before it could be considered as a tool for doctors, Google Health said.

"The sample was representative of practice at multiple sites of the worlds largest eye hospital, but more work is needed to understand the model performance in different demographics and clinical settings," it said.

"Such work should also understand the impact of unstudied factorssuch as additional imaging teststhat might be important for prediction, but were beyond the scope of this work."

The British Medical Journal recently warned that many studies claiming AI is superior or as good as human experts at interpreting medical images may exaggerate machines' diagnoses ability and pose a potential risk to patient safety.

It cautioned that overpromising language "leaves studies susceptible to being misinterpreted by the media and the public, and as a result the possible provision of inappropriate care that does not necessarily align with patients' best interests".

See original here:
DeepMind: Google AI can predict the progression of sight-threatening eye conditions - inews

This entry was posted on May 19, 2020 at 5:48 pm and is filed under Preventative Medicine. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed. |