← Back to Blog
ApproachSeptember 25, 2026
Why we want radiologists involved from the start
A model can score well on a benchmark and still be something no radiologist wants on their worklist. To avoid that, we intend to involve practicing radiologists early in pilot testing, instead of asking for a sign-off at the end.
We expect it to shape a lot of small decisions. How a finding is shown in the viewer. How a confidence score is worded. Which false positives are tolerable and which are not. How easily a radiologist can override a suggestion or dismiss it.
It also affects how we describe the product. We call the AI features 'planned' and 'early concept' because that is accurate today, not because it looks cautious on a website. When there is evidence to say more, we'll say more. Not before.