Bracco and Avicenna.AI have received US Food and Drug Administration 510(k) clearance for CINA-iCVA, a software tool designed to detect incidental intracranial aneurysms on CT imaging studies ordered for other reasons. The companies announced the clearance this week.
Incidental findings are a specific, well-defined problem. A patient receives a head CT after a fall; the scan is read for trauma, and a small aneurysm elsewhere in the image goes unnoticed because no one was looking for it. Software that flags such findings does not replace the radiologist’s primary read. It adds a second pass.
Why this clearance is a useful signal
The FDA’s 510(k) pathway for imaging AI has become more predictable over the past two years, but clearances still vary in clinical significance. Tools that target a narrow, measurable task with a clear failure mode tend to integrate into workflows more smoothly than broad diagnostic systems.
Aneurysm detection on CT fits that profile. Performance can be evaluated against ground-truth reads, false positives are manageable in a radiology workflow, and the consequence of a missed finding is severe enough that even modest sensitivity gains matter.
Integration is the harder part
Clearance is permission to sell, not evidence of adoption. Hospitals evaluate imaging AI on how it fits into picture archiving systems, how many additional alerts it generates, and whether radiologists trust it enough to change behavior. A tool that fires too often will be ignored; one that is too quiet will be forgotten.
The companies will need to publish real-world performance data, not just the validation sets used for clearance, if they want the software to become routine.
The broader pattern
Clinical AI is accumulating most quickly in imaging tasks with clear endpoints: fractures, hemorrhages, nodules, now incidental aneurysms. Each clearance is incremental. Together they describe a radiology workflow that is quietly becoming software-assisted by default, one narrow tool at a time.