CAPFED

Côte d’Azur–Québec Collaboration to Develop a Federated Medical Imaging Analytics Platform

Background

AI can only be deployed in clinical practice if large, structured medical imaging datasets annotated by experts are available. Yet access to such datasets remains one of the main barriers to progress. Regulatory requirements also make it difficult to share data across healthcare institutions, and even more so across national borders.
In 2026, Université Côte d’Azur and Université Laval (Québec) are celebrating ten years of strategic partnership. To mark this milestone, they have identified AI applied to medical imaging as a priority area for collaboration.

The Project

CAPFED aims to deploy PARADIM, a platform developed at Université Laval, within the Côte d’Azur Health Data Warehouse (EDS) operated by GCS CARES. PARADIM covers the full medical imaging research lifecycle:

• GDPR-compliant anonymization;
• quality control and expert annotation;
• AI model training and inference pipelines, with full traceability;
• orchestration of computational workloads on high-performance computing (HPC) infrastructure.

The platform will be made interoperable with Fed-BioMed (Inria). This federated learning solution enables models to be trained without requiring data to leave the institution where it is stored.
The project is structured around two main areas:

• interoperability between PARADIM and Fed-BioMed;
• deployment of PARADIM within the GCS CARES Health Data Warehouse, which encompasses Nice University Hospital (CHUN), Centre Antoine Lacassagne, and Lenval.