Thanks to thousands of training images, DeepMind’s AI can now analyse retinal scans to detect eye disease.
Artificial intelligence (AI) has already demonstrated that it’s the ideal assistive tool in the world of medical diagnoses, from heart disease to breast cancer.
This stems from its ability to detect small anomalies in scans, and other details that even an experienced doctor might miss. Pattern recognition abilities such as this are built via thousands – in some cases millions – of training images.
Thanks to Google’s DeepMind, AI’s augmentation of human expertise now extends to eye health, too. In partnership with Moorfields Eye Hospital in London, DeepMind has developed an AI that analyses retinal scans to spot early-stage eye diseases.
Part of the Alphabet group, DeepMind is a world leader in AI, led by Dr Demis Hassabis. It first announced the partnership with Moorfields, one the world’s leading ophthalmology hospitals, in 2016. Since then, doctors have been training the DeepMind algorithm with thousands of 3D retinal scans of healthy and diseased eyes.
DeepMind Health’s vision
The research could have a massive beneficial impact.
Two million people live with sight loss in the UK – three percent of the population – of whom around 360,000 are registered as blind or partially sighted. Glaucoma, diabetic retinopathy, and macular degeneration are the three major diseases responsible.
Faster, more reliable detection could make a huge difference in detecting problems and preserving a patient’s vision.
“In specific areas like medical imaging, you can see we’re going to make really tremendous progress in the next couple of years with artificial intelligence,” Dominic King, clinical lead for DeepMind Health, told the Financial Times. “Machine learning could have a very important role in picking up things more sensitively and specifically than currently happens.”
By leveraging Deep Leaning, the AI can detect minute, pixel-level trends and signals in the 3D scans, at levels of detail that medical professionals can’t hope to match, enabling them to make more informed decisions in indeterminate cases.
Internet of Business says
This isn’t DeepMind’s only foray into healthcare. The company is working with University College London Hospital to develop an algorithm that can identify cancerous tissue in the head and neck. However, the company faced controversy in 2016 when it signed a deal with the NHS to obtain patient data to test its Streams medical analytics app. The backlash led DeepMind to establish its own AI ethics board.
Nonetheless, its generalised AI has the potential to augment diagnostics across all fields of medical imaging, alongside the likes of Lunit and Ultromics. In the future, AI robotics and the IoT will all play vital roles in supporting ageing populations and stressed healthcare systems by making them more efficient and effective.
Keep up with the latest developments in our IoT in healthcare section.