AI Can Detect Signs Of Alzheimer's Years Before Symptoms Develop


Rosie McCall 18 Sep 2017, 22:05

Depending on the stories you read, artificial intelligence (AI) can be absolutely terrifying or downright hilarious (or both). Sometimes it can be lifesaving.

Scientists are constantly discovering new ways this exciting technology can be used to improve our health and extend lifespans, whether that’s software to detect skin cancer or a program that can pick the most viable embryo option for IVF treatment. Now, researchers are using AI scans to detect Alzheimer’s almost a decade earlier than doctors making a diagnosis based on symptoms alone.

Alzheimer's is the most common form of dementia, affecting over 5 million Americans. It's a progressive, neurodegenerative disease – patients might start off by feeling a little more forgetful or confused than usual, but their symptoms will worsen over time. Although there isn't a known cure, drugs and lifestyle changes can be effective at slowing down and reducing symptoms when the disease has been discovered early.

In a study, published earlier this month, researchers developed a machine-learning algorithm to detect Alzheimer's in brain scans 86 percent of the time. Even more impressively, it identified changes in the brain that showed mild cognitive impairment (MCI) 84 percent of the time. 

Nicola Amoroso, Marianna La Rocco, and colleagues from the University of Bari, Italy, taught AI software to tell the difference between healthy and unhealthy brains using MRI scans from the Alzheimer’s Disease Neuroimaging Initiative. Each scan was split into several small sections, which were analyzed for neuronal connectivity. The researchers discovered that the algorithm was most effective at analyzing brain regions of 2,250 to 3,200 cubic millimeters – which just so happens to be the same size as anatomical structures associated with the disease (e.g. the amygdala and hippocampus), La Rocca told New Scientist.

Full Article

If you liked this story, you'll love these

This website uses cookies

This website uses cookies to improve user experience. By continuing to use our website you consent to all cookies in accordance with our cookie policy.