A high-profile webinar titled “Harnessing AI for Clinical Decision-Making: Insights from Leading European Research Initiatives” brought together 230 participants from across the radiology and imaging fields yesterday. Organised by the European Society of Radiology (ESR) and the European Institute for Biomedical Imaging Research (EIBIR), the event showcased leading applications of artificial intelligence in medical imaging, with a focus on breast cancer diagnostics, interventional radiology, and data privacy.
The session was opened by EIBIR Scientific Coorinator, Prof. Regina Beets-Tan, Scientific Director of EIBIR, who introduced the webinar structure. The spotlight was on three major EU-funded initiatives – RadioVal, Odelia, and SHERPA – each demonstrating innovative approaches to AI-driven healthcare.
Oliver Diaz presented the RadioVal, which aims to develop and validate trustworthy AI models for breast cancer diagnosis and treatment planning using breast MRI and clinical data. He emphasised the Future-AI guidelines developed at the University of Barcelona and highlighted the importance of fairness, transparency, and explainability. The project includes eight international hospital partners and focuses on adaptability across diverse clinical settings.
Oliver Saldanha introduced ODELIA, a project leveraging swarm learning—a decentralised AI method that enables collaborative model training while maintaining patient data privacy. His talk highlighted promising early results in breast cancer diagnostics and demonstrated the value of privacy-preserving learning systems in healthcare.
Robert Hofsink presented the SHERPA, which aims to support decision-making and patient stratification in interventional radiology using AI and robotic assistance. The project addresses challenges such as clinician shortages and procedural complexity, with clinical trials planned across seven countries. SHERPA is funded by the Innovative Health Initiative and involves 16 partners from across Europe.
A dynamic panel discussion followed with audience questions submitted. Discussions centred on AI model performance, data requirements, and regulatory frameworks. All panellists emphasised the necessity of trustworthy, explainable, and clinically adaptable AI tools.
Prof. Beets-Tan concluded the event by reaffirming EIBIR’s commitment to supporting multi-centre AI research. She also announced plans for follow-up webinars to share further results and insights from the projects.