Connect with us

Breaking News

Doctors could team up with AI to spot dangerous brain bleeds faster


Scientist

Doctors could team up with AI to spot dangerous brain bleeds faster

By Clare Wilson It can sometimes be tricky to identify a brain haemorrhageZEPHYR/SCIENCE PHOTO LIBRARYAn AI that can spot a brain haemorrhage on an X-ray scan could help diagnose strokes, head injuries and ruptured blood vessels. The software was able to identify signs of bleeding in the head with similar accuracy to radiologists. While computers…

Doctors could team up with AI to spot dangerous brain bleeds faster

By Clare Wilson

Brain haemorrhage X-ray

It can sometimes be tricky to identify a brain haemorrhage

ZEPHYR/SCIENCE PHOTO LIBRARY

An AI that can spot a brain haemorrhage on an X-ray scan could help diagnose strokes, head injuries and ruptured blood vessels. The software was able to identify signs of bleeding in the head with similar accuracy to radiologists.

While computers won’t be replacing doctors any time soon, one area where they are making progress is in identifying signs of disease from images. Software can recognise moles that are likely to be cancerous from photographs, as well as eye damage caused by diabetes from pictures of the back of the eye, with near-human levels of ability.

Now Esther Yuh at the University of California, San Francisco, and her colleagues have developed a program that can interpret CT scans of heads. These involve multiple X-rays taken to create detailed images.

Advertisement

Did You See This CB Softwares?

37 SOFTWARE TOOLS... FOR $27!?

Join Affiliate Bots Right Away

If someone goes to hospital with symptoms suggesting brain damage, such as confusion or dizziness, they may have a CT scan of their head. But it can be hard for doctors to identify a tiny area of bleeding from the black and white images, says Yuh.

Her team first trained the software, called PatchFCN, on nearly 4400 head CT scans where the diagnosis was known. When they tested it on a new set of 200 randomly selected images, the AI performed similarly to four radiologists.

Yuh says the software could initially be used alongside doctors to speed up their work, but images would still need checking by a human. “No one’s just going to let it rip yet. We need to validate it in actual clinical practice to show it helps improve performance,” she says.

Journal reference: PNAS, DOI: 10.1073/pnas.1908021116

More on these topics:

Click to comment

You must be logged in to post a comment Login

Leave a Reply

To Top