AI fracture detection in adults and children, no age limit

Yes, and it is not Milvue saying so. An independent hospital team tested three commercial solutions on 1,500 consecutive emergency radiographs. TechCare Alert posted the highest diagnostic accuracy of the three, 90.1%, ahead of 88.8% and 71.0% (Academic Radiology, 2023). Detection is validated with no age limit across Europe, from infant to adult; the FDA clearance (510(k) K242171) adds an AUC of 0.962 in fracture detection, standalone. The decision stays with the radiologist. Full details on the Detection / TechCare Alert page.

Three independent studies, all pointing the same way. A multivendor study in Zurich (Pediatric Radiology, 2025) put Milvue ahead on the hardest pediatric reads, the elbow and the joint effusion, with the lowest rate of doubt. A second (Pediatric Radiology, 2025) measured the time saved on fractures with high medico-legal stakes, without giving up accuracy. External validation on the elbow (Diagnostic & Interventional Imaging, 2023) confirms robustness in children. The common thread: external evidence, never self-proclaimed.

As a second read, the AI is a safety net, not a judge. The FDA reader study puts a number on it: for adult fracture, the radiologist's sensitivity rises from 0.81 to 0.98 once assisted (aided vs unaided). This is not a ranking between vendors; it is what the eye catches at the end of a shift, when fatigue sets in. Detection covers 7 pathology families in a single pass, and the radiologist makes the final call on every image.

In the ED, the AI second-reads every radiograph, every time. In a study combining MSK and chest detection (European Journal of Radiology Open, 2023, 1,772 patients), it reaches an overall AUC of 0.954; more importantly, emergency physicians misread nearly one case in ten, and the AI catches 90% of those. At Aalborg University Hospital (Denmark), prioritization helped deliver up to 1 hour less waiting. The pre-report lands in under 30 seconds. AI does not make errors disappear: it catches them where they hide.

Very fast, with no change to how you work. The pre-report is available in under 30 seconds after the image, and the full analysis returns to the PACS in under 2 minutes. The solution relies on DICOM and HL7 standards and on the IHE AIR (AI Results) interoperability profile, for native integration into the RIS/PACS workflow, with no second screen or manual export. The technical detail lives in our RIS/PACS integration guide.

The Milvue Suite is a class IIa medical device (EU MDR 2017/745, notified body BSI CE 2797); in the US, TechCare Trauma is FDA cleared (510(k) K242171) and SmartTrauma is licensed by Health Canada. Three regulatory zones, one detection engine. On that foundation, one rule never changes: Our AI does not replace radiologists. It makes them irreplaceable.