At an event organized by the European Commission, the European Society of Radiology, and the EUCAIM (European Cancer Imaging Initiative) project on 6 February in Brussels, Belgium, Luis Martí-Bonmatí presented how combining medical imaging with AI technologies can revolutionize cancer research and treatment.
Martí-Bonmatí, Director of the Medical Imaging Department and Chairman of Radiology at La Fe University and Polytechnic Hospital in Valencia, Spain, illustrated the relevance of medical imaging for the diagnosis and treatment of cancer using a case: an apparently healthy patient suddenly suffered a seizure. After a CT and MRI scan, a glioblastoma was diagnosed. Martí-Bonmatí made it clear that many of these tumors recur despite treatment. In addition, there are areas of the brain that are difficult to detect using conventional imaging, but which play a decisive role in tumor spread. Genetic and molecular changes, crucial for personalized therapy, also often remain undetected.
Martí-Bonmatí: “Computer-aided, quantitative image analysis can help to overcome these challenges. Radiomic features can be linked to genetic or molecular changes. These markers are crucial for developing personalized treatments.” However, he also emphasized that there has been a major problem in the past with the lack of reproducibility of such radiomic analyses. The reasons for this are manifold: images come from different sources, with different devices, recording protocols, and methods, which leads to deviations in the results. This is where AI comes in: It can harmonize image data, recognize patterns, link imaging information with clinical and molecular data, and thus create more precise diagnosis and treatment options.
EUCAIM: pan-European federated infrastructure
The EUCAIM initiative – a pan-European federated infrastructure that combines medical imaging with AI technologies – was launched to make the most of the opportunities offered by AI. The aim is to create a comprehensive platform that links data from different hospitals and research institutions and makes it available to develop AI-supported applications. To this end, a decentralized network has been set up in which hospitals, national networks, and research repositories can make their data available. This is done in accordance with the highest data protection and security standards so that sensitive health data can be processed securely.
42,000 patient data collected
In the first two years of the EUCAIM initiative, a central platform was put into operation with reference sites in Valencia and Rotterdam. A total of 57 data collections with over 42,000 patient data records were compiled. The platform enables decentralized data analysis so that AI models can be trained without having to store sensitive patient data centrally. It is increasingly being used by researchers, clinics, and companies to develop new AI-supported solutions. Currently, 94 partners and over 175 stakeholders are already involved. The EUCAIM platform is not only used to store and analyze data but is also used for specific research projects. One example is the Leopard project, which deals with liver cancer and liver transplants. Here, AI methods are used to extract radiomic features from medical image data that help to develop personalized treatment strategies.
Another important goal is the validation of AI algorithms in accordance with European regulatory requirements (MDR, GDPR). This should not only support research but also help companies to bring AI-supported medical products to market maturity.
Platform to be further expanded
Finally, Martí-Bonmatí emphasizes that the platform will continue to grow. The next steps involve integrating additional data sources and hospitals while also developing new AI models for other types of cancer. A standardized, legal, and ethical framework for data exchange will also be established.
