Hospitals train scan-reading software together without sharing patient scans
Original title: Performance changes in automated lesion detection under federated learning with sequential institution addition.
How far along is this research?
- Lab cells
- Animals
- Review
- Tested in people
This was done on cells in a lab, not in people. It is a very early step.
This was a computer test of software, not a treatment tried in people.
The short version
This study tested a way for hospitals to build better scan-reading software as a team.
What was studied. Researchers used two types of computer software that look for problems on brain scans. One looked for weak spots in brain blood vessels. The other looked for tumors that had spread to the brain. Scans from each hospital were added to the training one hospital at a time.
What they found. Adding hospitals one at a time worked better than starting the training over from nothing. Updating only part of the software worked about as well as updating all of it. It also needed far fewer parts to be changed. Adding all the hospitals at the same time gave less steady gains.
What this means, and what it doesn't
What it could mean: Software like this may help doctors spot things on brain scans. Hospitals could help train it while keeping their own scans in house. This study is about the software itself, not about a new treatment.
What it doesn't mean: This is not a treatment, and it is not a cure. It was a computer test that used scans already collected. No one was treated in this study. Nothing here changes the care you get today.
Source: PubMed, September 27, 2026 · Read the original
This plain-language summary was written by AI and published automatically after passing our automatic safety checks. How we write.