Implementation of an interoperable pan-European federated repository of clinical data and medical images, including secure data sharing mechanisms

An Interoperable pan-European federated repository of clinical data and medical images that allows sharing data in compliance with legal, ethical, privacy and security requirements, for AI-related training and experimentation; the repository will rely on federated data storage and will operated on a Federated Learning basis, abiding to the highest data privacy and security standards. It will also offer High Performance Computing-as-a-service, where necessary, thereby allowing for cost-effective performance of computationally intensive processing without the need for maintaining expensive equipment.


Interested in sharing medical images and/or clinical data to contribute to INCISIVE’s repository? Register your interest and we’ll be in touch with more details.


These are the INCISIVE Goals:

  • Implement an interoperable data integration service: based on a common data model and well-established standards (HL7 FHIR, DICOM, SNOMED, LOINC), enabling data providers to easily link their data sources to the INCISIVE platform.

  • To incorporate a federated data storage schema: enabling the data providers to keep full control of their data, ensure their autonomy and at the same time contribute to the platform.

  • To implement a secure data exchange/sharing and privacy mechanism: based on blockchain technology and eIDAS, enabling secure access to data and identity management and ensuring trust.

  • To develop an automated data curation and an ML-based annotation service: ensuring high- data quality, correct interpretation of data/results, increasing speed and efficiency of annotation by minimizing the human intervention.

  • To define and implement a data donorship mechanism: following all related legal directives and enabling users to share their data while keeping full control of it.

  • To incorporate a data anonymization mechanism: enabling data processing, with respect to all related legal and ethical aspects. The mechanism will be bi-directional, including also a de-anonymization process.

The INCISIVE project decided that the data upload process will be carried out using an Excel file that indicates the terms to be reported and that each Data Provider will upload to their Federated Node or in the Central Node anonymously indicating a patientID. This mechanism makes it possible to send data from different patients from the same data provider with the same file. Four file templates have been generated, one for each cancer type with different clinical data, for this reason it was decided to create four different HL7 FHIR messages, one for each cancer type.

INCISIVE clinical data was semantically encoded in a way that gets everyone within a system to speak the same language and understand the meaning of the data. Templates contain input fields about clinical elements and laboratory elements; it was distinguished between these two types of data and used SNOMED CT or LOINC according to the medical concepts.

With DICOM and HL7 FHIR, INCISIVE achieves a Common Data Model (CDM) that allows receiving, storing, and processing information from multiple sources, multiple data providers, as a one standard way.

For more detail, see INCISIVE Standardization Suggestions, and the HL7 FHIR INCISIVE Implementation Guide.

Take a look to the recommendations of interoperability standards for projects managing clinical data and medical images.