Table 3 Metadata management approaches of the AI4HI projects
From: Position of the AI for Health Imaging (AI4HI) network on metadata models for imaging biobanks
Project | Metadata collection | Metadata types | Models used | Unique characteristics |
|---|---|---|---|---|
PRIMAGE | Structured e-forms | Imaging, clinical, image radiomic analysis | DICOM for imaging metadata MIABIS for biological samples and tissue OMOP-CDM for clinical | Integration of the DICOM and MIABIS standards, and metadata model that captures the biomechanical/signalling behaviour of tumours |
EuCanImage | Structured e-forms | Imaging, clinical | DICOM-MIABIS for imaging data Extension of ICGC-ARGO for clinical variables | Link between imaging and non-imaging data |
INCISIVE | Structured e-forms | Clinical, biological, imaging | Multiple terminologies for clinical data (e.g., SNOMED-CT, ICD10, ATC classification) FHIR for communication | Data Integration Quality Check Tool employed to identify whether data follow the harmonisation requirements defined |
CHAIMELEON | Structured e-forms | Imaging, clinical | DICOM for imaging metadata MIABIS for biological samples and tissue OMOP-CDM for clinical | A multimodal analytical data engine will facilitate interpretation, extraction, data harmonisation, and exploitation of the stored information. The CHAIMELEON repository will ensure the usability and performance of the repository as a tool fostering AI experimentation |
ProCancer-I | Data upload tool (e-forms) | Imaging, clinical | DICOM-Radiation therapy for imaging data OMOP-CDM for clinical data | Provides an extension to OMOP-CDM going beyond radiology/oncology extensions and introduces another model (AI passport) for modeling analysis workflows and AI development |