Seminal
The magazine and atlas of EU-funded researchA new publication from the web and AI team at Scott Charnley

Atlas / Germany

European Research Council grant · Running

RAD-AI-INFRA From Research Infrastructure to Radiology AI Platform

Coordinator
Universitatsklinikum Bonn, Germany
Runs
1 October 2026 to 31 March 2028
Status
Running
EU funding
€150k
Total cost
—
Partners
1 organisations in 1 countries
Type
European Research Council grant
Grant agreement
101335506

What the project is doing

In the project's own words, from its summary on CORDIS.

Radiology AI faces a dual infrastructure crisis: developing next-generation AI, particularly radiology foundation models, requires access to large-scale clinical imaging data that remains siloed across hospital systems. Evaluating the growing number of commercial AI products requires the same data access, yet hospitals cannot benchmark these tools on their own patients before procurement, cannot identify performance gaps, or systematically monitor performance after deployment.

We developed and open-sourced two on-premise infrastructure components that address this bottleneck: ADIT (Automated DICOM Transfer), for vendor-agnostic retrieval of DICOM imaging data from hospital archives; and RADIS (Radiology Report Archive and Discovery System), which transforms unstructured radiology reports into a structured, searchable resource. Both tools were substantially developed within the ERC Consolidator Grant AI-NEXT and deployed at 2 major German university hospitals with different PACS/RIS vendors, indexing multi-million scan/report pairs.

The platform enables two complementary use cases: first, large-scale, GDPR-compliant data discovery and export for AI research; second, independent evaluation of (commercial) radiology AI products on local clinical data, from benchmarking and gap analysis through fine-tuning to post-deployment monitoring under the EU AI Act. This project will mature ADIT and RADIS into a unified, deployment-ready platform through technical hardening, interoperability engineering across heterogeneous hospital IT environments, pilot deployments at external sites, and preparation of a commercial exploitation pathway.

RAD-AI-INFRA will provide a governance-compatible radiology AI infrastructure targeting 50–100 European academic centres initially, with scaling potential as AI Act and EHDS compliance drive demand across the EU's 15,000+ hospitals and radiology practices, as well as AI vendors requiring access to data & infrastructure for real-world validation.

Topics

Who is involved

OrganisationCityCountryRole
Universitatsklinikum BonnBonnGermanyCoordinator

Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI