Helping clinicians look twice at images
Could AI flag small findings that are easy to miss?
Reading an image and deciding what it means is the clinician's job, and we cannot take that over. We would like to study a tool that marks easy-to-miss areas along with its evidence, so a clinician can take one more look. We want to be open about how we build it and what we learn.
CT, MRI and X-ray images
We aim to connect and clean the images, then have agents locate regions of interest and verify each other.
Marked regions with evidence; the final judgment stays with the clinician
| Field | Medicine |
|---|---|
| Status | Building pipeline |
| Input | CT · MRI · X-ray |
| Output | Marked regions · evidence (for clinician review) |
| Delivery | Joint research |