Skip to content
Contact us

How the EuCanImage Project is Transforming Cancer Imaging with AI and Federated Data Platforms in Europe

eucan image case study image (1)

Industry

Healthcare

Challenge

Currently, nearly all cancer treatments are guided by human expertise and medical images.These images are typically stored locally in each hospital.Central image collections of past exams with open access generally don't exist yet.

Results

A centralized repository of all image data with standardized annotation protocols established a cohesive workflow across European hospitals. The platform's consistent processing methodology and comprehensive annotator guidelines minimized intervariability, regardless of image origin or source data characteristics.

Key Product

Research

eucanimage.eu/
20 000
Single Subjects
20
partners
11
countries
4
years

EuCanImage

  • UNIVERSITAT DE BARCELONA (Coordinator), Spain
  • UNIVERSITEIT MAASTRICHT, Netherlands
  • ERASMUS UNIVERSITAIR MEDISCH CENTRUM ROTTERDAM, Netherlands
  • BARCELONA SUPERCOMPUTING CENTER – CENTRO NACIONAL DE SUPERCOMPUTACION, Spain
  • FUNDACIO CENTRE DE REGULACIO GENOMICA, Spain
  • UNIVERSITY OF ARKANSAS FOR MEDICAL SCIENCES, United States of America
  • BIOBANKS AND BIOMOLECULAR RESOURCES RESEARCH
  • INFRASTRUCTURE CONSORTIUM (BBMRI-ERIC), Austria/ 3rd Party of BBMRI: Technologie- und Methodenplattform für die vernetzte medizinische Forschung e.V. (TMF), Germany
  • UNIVERSIDAD DEL PAIS VASCO/ EUSKAL HERRIKO UNIBERTSITATEA, Spain
  • LYNKEUS, Italy
  • COLLECTIVE MINDS RADIOLOGY AB, Sweden
  • RADIOMICS, Belgium
  • SIEMENS HEALTHCARE GMBH, Germany
  • EIBIR GEMEINNUTZIGE GMBH ZUR FORDERUNG DER ERFORSCHUNG DER BIOMEDIZINISCHEN BILDGEBUNG, Austria
  • EUROPEAN SOCIETY OF ONCOLOGIC IMAGING ESOI EUROPAISCHE GESELLSCHAFT FUR ONKOLOGISCHE BILDGEBUNG, Austria
  • EUROPEAN ASSOCIATION FOR CANCER RESEARCH, United Kingdom
  • UNIVERSITA DI PISA, Italy
  • FUNDACIO CLINIC PER A LA RECERCA BIOMEDICA, Spain
  • UMEA UNIVERSITET, Sweden
  • GDANSKI UNIWERSYTET MEDYCZNY, Poland
  • LIETUVOS SVEIKATOS MOKSLU UNIVERSITETO LIGONINE KAUNO KLINIKOS, Lithuania

Key Barriers to Advancing AI in Cancer Imaging

Currently, nearly all cancer treatments are guided by human expertise and medical images.
These images are typically stored locally in each hospital.
Central image collections of past exams with open access generally don't exist yet.

Several significant challenges were identified in the realm of cancer imaging and artificial intelligence (AI) in oncology:

  • Lack of Large-Scale, Multi-Center Cancer Imaging Repositories: There was a notable absence of extensive, high-quality cancer imaging datasets in Europe. This scarcity hindered the development and validation of AI models, as existing datasets were often limited to single institutions, reducing the generalizability and robustness of AI solutions across diverse populations and clinical settings.
       
  • Data Annotation Complexities: Preparing imaging data for AI applications required meticulous processes, including extraction from hospital records, anonymization, and detailed annotation. This task was labor-intensive, necessitating radiologists to identify and label tumors and lesions accurately. The lack of standardized annotation protocols further complicated the integration of data from multiple sources.

  • Fragmented Clinical Systems and Legal Frameworks: Europe's diverse clinical systems and varying legal regulations posed challenges in harmonizing data collection, sharing, and analysis. The absence of unified standards and interoperable systems impeded collaborative efforts and the establishment of a cohesive cancer imaging platform.

  • Integration of Multi-Scale Data: Combining imaging data with other biological and health data (such as genomic, molecular, and clinical information) was essential for developing comprehensive AI models. However, integrating these diverse data types into a cohesive platform presented technical and methodological challenges.

    Addressing these challenges was imperative for advancing AI-driven oncology research and improving personalized cancer care across Europe.

EUCanImage: Building Europe’s Federated AI-Powered Cancer Imaging Platform

Imagine what we could achieve by combining large image collections with artificial intelligence.
The goal of the EUCanImage Project is to build a highly secure, large-scale, and federated European cancer imaging platform.
This platform will have capabilities that greatly enhance the potential of artificial intelligence in oncology.

The EUCanImage Platform will include 25,000 new data points for research, aimed at improving the detection of small liver tumors and metastases of colorectal cancer, as well as estimating subtypes of breast tumors for treatment planning.

The platform will be connected to existing biological and health databases.
This enables the development of AI solutions that integrate different types of data, including genetic, molecular, and biochemical information, into dense, patient-specific cancer fingerprints.

To deliver this platform, the project will connect to other research infrastructures and cloud computing capabilities.
It will build upon several key European initiatives in high-quality data sharing for personalized medicine research, including Euro-BioImaging and the European Genome-Phenome Archive.

The project is also collaborating with the Cancer Imaging Archive in the US to leverage their years-long experience in cancer imaging storage, curation, and management.

World-renowned experts in cancer research, AI, and bioethics are working together to establish necessary guidelines for developing standardized, trusted, and transferable decision-support systems for future clinical oncology.

Impact and Future Potential of the EUCanImage Cancer Imaging Ecosystem

The EUCanImage initiative is currently developing a GDPR-compliant, scalable cancer imaging platform that demonstrates the feasibility of linking high-quality, large-scale medical imaging with biological and health data.

The expected outcomes include:

Validated AI Decision Support Tools for Precision Oncology: By integrating diverse data types and ensuring clinical-grade reliability, the platform will promote AI solutions that are both effective and trusted by medical professionals.

Enhanced Clinical Adoption Across Europe: The project sets the stage for widespread implementation of AI-powered tools in oncology, fostering a collaborative research environment and improving personalized treatment outcomes.