01Integration & interoperability

Integrating radiology AI into the RIS/PACS

How to connect imaging AI to the RIS/PACS through DICOM, HL7, the worklist and IHE profiles, without changing how the radiologist works.

Medically reviewed by Dr Alexandre Parpaleix, MD-PhD, CEO and co-founder of Milvue

How does radiology AI integrate with the RIS/PACS via DICOM and HL7?

The AI fits into the existing workflow: the PACS sends it images in DICOM, it returns results to the PACS, and status flows to the RIS over HL7. No new tool to open. The Milvue Suite detects 7 pathology families in a single pass and returns the full analysis to the PACS in under 2 minutes, with a pre-report ready in under 30 seconds. The radiologist stays in charge: the AI assists, it does not replace anyone.

What does AI change for the radiologist's worklist?

Results appear directly in the worklist, where the radiologist already works, with no software switch. Because the results land where the work happens, the gain is measurable: in an image-to-report workflow, TechCare Report integrated in the RIS/PACS delivered up to +64% more exams read per session (figure reported by a partner radiology center). TechCare Report is intelligent radiology dictation, subscription software, not a medical device.

Why do IHE profiles matter for deployment?

IHE profiles ensure the AI talks to the RIS/PACS through proven standards rather than brittle custom connectors. Milvue relies on DICOM and HL7 interoperability and on the IHE AIR (AI Results) profile, the one that standardizes how AI results are returned. In practice, integration risk drops, and a RIS or PACS update does not break the connector.

Do you have to replace your RIS/PACS to add AI?

No. The AI plugs into existing infrastructure through integrators and marketplaces, without replacing the RIS or PACS. Milvue is available via AI marketplaces (Incepto, Blackford, deepc) and the main RIS/PACS integrators. This modular approach activates the 7 pathology families and up to 60 automatic MSK measurements (Cobb, full legs, EOS, bone age) without disrupting the clinical workflow. You add a capability, you don't redo the plumbing.

What clinical benefit follows once the AI is integrated?

Beyond productivity, integration supports diagnostic quality. In an ED study combining MSK and chest detection (European Journal of Radiology Open, 2023, 1,772 patients, overall AUC 0.954), the AI catches 90% of the cases an emergency physician had misread; at Aalborg University Hospital (Denmark), prioritization helped deliver up to 1 hour less waiting. In fracture detection, Milvue posts the highest accuracy in the only published independent comparative study. The technology serves the read, not the other way around.

Is the integration compatible with pediatric use?

Yes. Fracture detection is validated with no age limit in Europe, in adults and children alike. External validation for pediatric elbow fractures and effusions is published in Diagnostic & Interventional Imaging (2023). The same RIS/PACS pipeline serves adult and pediatric cases, with no separate configuration. For the clinical detail, see our fracture detection guide.

The Milvue solution The Milvue Suite
Explore

See Milvue AI in real conditions.