Building an Ontology-Based Cohort of Liver Cancer Imaging Data for AI Development on the European Federated Platform EUCAIM.

Hepatocellular carcinoma (HCC) is steadily increasing in incidence worldwide and requires data-driven approaches to improve diagnosis, prognosis, and therapeutic decisions. We describe the harmonization of IMALIVE -a real-world HCC cohort- with the common data model of the European Cancer Imaging Initiative (EUCAIM). IMALIVE integrates demographic, clinical, and imaging-related variables, which were aligned with core variables of the EUCAIM common data model through a robust mapping process and expert validation. The process achieved full coverage of the core dataset and added HCC-specific variables, including tumor staging and liver function scores. In addition, metadata from digital pathology was standardized using the international MIABIS/BBMRI model, extending interoperability across radiology and histology. This work demonstrates the feasibility of harmonizing local cohorts enabling their integration in the data catalogue of large scale federated platforms It highlights how medical experts' contributions to this harmonization process can enrich common models with clinically relevant variables for AI development in their domain of expertise.