AI can detect Alzheimer's 10 years before symptoms show up

Leslie Hanson
September 18, 2017

Medical researchers have found artificial intelligence as an unlikely but very much welcome ally in the development of Alzheimer's disease detection techniques. The algorithm has been developed by researchers Nicola Amoroso and Marianna La Rocca from the University of Bari, Italy, and has proved highly successful during initial tests.

They developed an algorithm that can spot tiny structural changes in the brain caused by the disease a decade before symptoms even appear. Then, they divided these scans into small regions and allowed the system to analyse the neuronal connectivity between them.

The team from the University of Bari trained their AI by feeding it 67 MRI scans, 38 of which were from patients suffering from the disease and 29 MRI scans of people that do not have it.

The team then tested the algorithm on a second set of scans from 148 subjects. The AI was able to correctly detect Alzheimer's, with a success rate of 86% - and it could also identify "mild cognitive impairment" 84% of the time. Researchers were limited to the scans in USC LA's Alzheimer's Disease Neuroimaging Initiative database.

"Cerebrospinal fluid analyses and imaging using radioactive tracers can tell us to what extent the brain is covered with plaques and tangles, and are able to predict relatively accurately who is at high risk of developing Alzheimer's 10 years later", a researcher said. After all, early detection gives people the power seek treatment that can slow down the condition's effects, as well as enough time to get their legal and financial affairs in order. The algorithm is far from ideal, but with added training from more MRI scans, it will likely develop into a valuable tool for early Alzheimer's disease detection.

This isn't the first study that involves using "cutting-edge technology to help diagnose Alzheimer's and other neurodegenerative diseases", reports Digital Trends.

Other reports by Iphone Fresh

Discuss This Article