Imaging AI spots and predicts Alzheimer’s signs 6 years early in PET scan study

In Silico
The AI was able to better track the more subtle, distributed changes in brain metabolism seen in Alzheimer's patients. (Pixabay / Geralt)

Using a deep learning algorithm and PET images of the brain, researchers were able to train an artificial intelligence program to spot the signs of Alzheimer’s disease more than six years ahead of a final clinical diagnosis.

Researchers at the University of California, San Francisco, scanned for small changes in the metabolism of brain cells by tracking radioactively tagged glucose molecules, or fluorine 18 fluorodeoxyglucose, across more than 2,100 images from about 1,000 patients.

The algorithm was able to predict every case that progressed to Alzheimer’s disease—with 82% specificity and 100% sensitivity—an average of 75.8 months early compared to radiologists.

Your Daily Newsletter — Free

Enjoying this story? Subscribe to FierceBiotech!

Biopharma is a fast-growing world where big ideas come along every day. Our subscribers rely on FierceBiotech as their must-read source for the latest news, analysis and data in the world of biotech and pharma R&D. To read on the go, sign up today to get biotech news and updates delivered right to your inbox!

“Differences in the pattern of glucose uptake in the brain are very subtle and diffuse,” said study co-author Jae Ho Sohn, from the UCSF’s Radiology & Biomedical Imaging Department. The study was published in the journal Radiology.

“People are good at finding specific biomarkers of disease, but metabolic changes represent a more global and subtle process,” Sohn added.

The AI was trained on 90% of the images, collected from the Alzheimer’s Disease Neuroimaging Initiative, before being tested on the remaining 10% plus an independent set of exams from 40 patients it had never studied. In addition, a heatmap of the deep learning model’s processes showed that it analyzed the entire brain as well as previously known areas of interest.

The team, including researchers from UC Berkeley and Davis, aims to use the AI in the future to recognize patterns in the accumulations of beta amyloid plaque and tau proteins.

While it still needs to be validated with a larger independent dataset and a prospective study, Sohn said he hopes the algorithm could one day complement the work of radiologists and other biochemical and imaging diagnostics.

“If we diagnose Alzheimer’s disease when all the symptoms have manifested, the brain volume loss is so significant that it’s too late to intervene,” he said. “If we can detect it earlier, that’s an opportunity for investigators to potentially find better ways to slow down or even halt the disease process.”