Imagine all the data - SPHN introduces medical imaging concepts for its data model

Over the years, a wealth of imaging data has accumulated in hospitals and imaging facilities. This is due to refinement in imaging techniques and devices and increased availability for patients. However, the FAIR use, reuse, and exploitation of these invaluable data resources remains a challenge.

A project that paved the way to represent imaging data in a structured way was the PHRT project Towards Deep Medical Imaging Biobanks (DMIB) which started at the Biomedical Image Computing Group of the ETH Zurich. Led by Edwin ter Voert (ETH Zürich / University of Zürich), the project gathered input from multiple clinicians and other medical imaging experts which was forged into new concepts representing imaging data as well as metadata.

In collaboration with the SPHN Data Coordination Center, these concepts have been refined and aligned with the SPHN data model. This project-specific extension, based on the SPHN Schema release 2025.1, now covers further important aspects of healthcare data including imaging methods, medical imaging datasets, use of pseudonymized metadata, relevant reagents, devices, and data standards (e.g., DICOM).

This improved framework will facilitate interaction with imaging-related projects including the Swiss Digital Pathology Initiative (SDPI) and the Biomedical Informatics Imaging (BMI2) project of The Loop Zurich. Now imaging data can be represented in a structured way, ready to fuel next-generation machine learning and AI technologies for leveraging the rich information medical images contain.

The SPHN Medical Imaging concepts are available in a project-specific dataset and schema for feedback and testing in the Medical Imaging Git repository. The medical imaging extension can also be explored interactively using SPHN Schema Scope, the interactive visualization component of the SPHN Metadata Catalog.