Skip to main content

The European Institute for Biomedical Imaging Research (EIBIR) presented two of its flagship projects under the Euro Beating Cancer Plan in a dedicated webinar earlier this month: the EUropean Federation for CAncer IMages (EUCAIM), an initiative that focuses on harnessing AI power in cancer imaging, and the SOLACE Project, to strengthen lung cancer screening in Europe. Mélisande Rouger reports the highlights of the event.

‘We are pleased to help with project management so that researchers can free their time to focus on science,’ EIBIR Scientific Director Regina Beets Tan from the Netherlands Cancer Institute (NKI) in Amsterdam said as she kickstarted the event. ‘EIBIR has helped to secure researchers almost 180 million EUR funding since its inception in 2006, and it is now run by over 489 partners around the world.’

One goal of the Euro Beating Cancer Plan is to make the most of the potential of data, including medical imaging data. Cancer imaging data sets and lakes already exist for many different types of cancer, but they are spread, scattered, and not easily accessible, according to EUCAIM coordinator Luis Martí-Bonmatí from La Fe University and Polytechnical Hospital in Valencia, Spain.

Harnessing the Power of AI

‘We need access to big data if we really want to make use of the power of AI, such as convolutional networks, and generative AI for the best of patients with cancer’ he said.

EUCAIM wants to foster innovation and deployment of digital technologies, and link resources and databases to establish an open infrastructure that will be both centralized and distributed, with the development and benchmarking of AI tools for prediction models.

‘We really want to showcase access and use of medical images while taking care of GDPR,’ Martí-Bonmatí explained. ‘We also want to establish a large amount of cancer images and link it with clinical and molecular data, in line with the European data space.’ 

Over the last two decades, researchers have gone deeply into signal analysis and digital computerization to extract radiomics data. But once they want to transfer what is working at the individual hospital level to other hospitals, one must recognize that most of those public resources did not translate into clinical practice, Martí-Bonmatí believes.

‘The reason may be related to the reproducibility of what we are obtaining and to the explainability and completeness of the approach,’ he said. ‘In the last few years, we are deeply thinking that it is not only medical imaging plus maybe radiomics, but also linking that information with the relevant molecular, analytical, clinical, pathological or genetic information we can obtain from specific patients on this integrative approach the one that will be pacing centre ability to link this multi-source data to estimate a clinical end point with accuracy.’ 

Regina Beets-Tan introducing Luis Martí-Bonmatí in an event organized by EIBIR on May 9th

Researchers need AI and digital tools, but also research communities and projects such as CHAIMELEON, where teams are able to produce some knowledge on how AI can be generated and images harmonized.

‘Our vision is to enhance our cancer diagnosis and treatment through AI tools based on medical imaging,’ he said. ‘To do that, we already deployed a hybrid federated infrastructure with a central structured repository to power up AI and imaging to beat cancer.’

The project will also provide a research platform for the development and benchmarking of AI tools and precision medicine; build an atlas of cancer images with millions of images from all imaging modalities; and create a federated data warehouse within the European health data space for development and observational studies.

With 21 clinical sites in 12 countries, EUCAIM has just submitted an Open Call to further unlock the power of AI against cancer. ‘I encourage all of you to help us through the webpage of EUCAIM to enlarge our community of partners,’ he concluded.

Implementing Nationwide Lung Cancer Screening Programs

In the next talk, Hans-Ulrich Kauczor from Heidelberg University in Germany presented the SOLACE project, which focuses on the implementation of low-dose CT lung cancer screening (LCS) programs within the EU.

‘We have evidence that low-dose CT screening for lung cancer will reduce mortality by 20 to 25%,’ he said. ‘The US have implemented an LCS program and there are multiple projects in Europe, but Croatia is the first country to have such a nationwide program. We are building upon their experience to improve the quality of LCS across Europe, and to make sure we can improve accessibility, benefit-harm balance, and cost effectiveness of these programs.’

SOLACE brings together for the first time all the stakeholders – radiologists, epidemiologists, pulmonologists, thoracic surgeons and oncologists – needed to design, plan and implement LCS on a regional and national level on a sustainable basis. The initiative provides a tool box of individualized approaches to LCS to overcome and identify bottlenecks and address health inequalities in different European countries.

After one year of existence, SOLACE has set up pilot projects that are run in ten countries, along separate work packages targeting for example project management (run by EIBIR); dissemination and outreach; and provision of papers and documents.

One of SOLACE’s goals is to reach more women as well as vulnerable, marginalized populations.

One pilot is focused on the needs of women, who may benefit more from using low dose CT LCS than men, Kauczor explained. ‘Looking at histology a subtypes, women are at higher risk or proportion of developing certain types of adenocarcinoma. Intermediately fast growing adenocarcinomas are best detected by the one year interval of low-dose CT screening, so women would benefit the most.’

However, in numbers, women are participating less in screening programs than men. ‘We have certain programs that increase participation of women in LCS and one activity is how we can liaise breast cancer screening with LCS run in ten different countries,’ he concluded.