The tool was trained by a team of University of California scientists using thousands of slide images of tissue taken from autopsies of both healthy and diseased brains. The researchers built a convolutional neural network out of the images to teach the program to recognize the differences between different types of amyloid plaques and to do so in biologically valid methods, rather than being based simply on programmed patterns.
As a result, the tool was able to process a slide depicting a whole-brain slice with 98.7 percent accuracy, which the study’s authors note does not supersede the abilities of a human neuropathologist.
“It’s a copilot, a force multiplier that extends the scope of what we can accomplish and lets us ask questions we never would have attempted manually,” said Michael J. Keiser, PhD, an assistant professor at UC San Francisco. “For example, we can look for rare plaques in unexpected places that could give us important clues about the course of the disease.”
More articles about AI:
Elon Musk’s neurotech startup falls short in funding round
FDA clears AI tool to flag pulmonary embolism in CT scans
3 new AI-enabled FDA ‘Breakthrough Devices’: an Alzheimer’s-detecting retinal scanner, a kidney disease diagnostic & more
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.