EuCanImage Webinar
Obstacles and avenues for data sharing and AI in cancer imaging
Wednesday, November 17, 2021
14:00 – 18:00 CET

Thank you

We would like to thank the speakers and moderators of the first EuCanImage webinar as well as the attendees who tuned in today.
It was a pleasure having you online and we hope to be able to welcoming you to the next webinar!

Programme

EuCanImage Webinar

Obstacles and avenues for data sharing and AI in cancer imaging: EuCanImage ESOI/EACR Webinar

 

14:00-14:10

Welcome by ESOI/EACR

Alberto Bardelli, Candiolo/IT
Emanuele Neri, Pisa/IT


14:10-14:20

Introduction to EuCanImage

Karim Lekandir, Barcelona/ES


14:20-14:30

Presentation 1 – CARING
Ethical and legal governance for data sharing & AI in cancer imaging

Moderation: Emanuele Neri, Pisa/IT
Davide Zaccagnini, Boston/US

 

14:30-15:10

Panel Session 1

Magdalena Kogut-Czarkowska, Brussels/BE
Marius Mayerhoefer, New York/US
Emanuele Neri, Pisa/IT
Davide Zaccagnini, Boston/US

 


15:10-15:20

Presentation 2 – HANDLING
Data anonymization and curation across clinical centres

Moderation: Karine Seymour, Toulouse/FR
Lawrence Tarbox, Arkensas/US

 

15:20-16:00

Panel Session 2

Haridimos Kondylakis, Heraklion/GR
Nickolas Papanikolaou, Lisbon/PT
Karine Seymour, Toulouse/FR
Lawrence Tarbox, Arkensas/US

 


16:00-16:15

Coffee Break

 


16:15-16:25

Presentation 3 – ANNOTATION
Obstacles and solutions for efficient and standardized data annotation in cancer imaging

Moderation: Elisa Oricchio, Lausanne/CH
Anders Nordell, Stockholm/SE

 

16:25-17:05

Panel Session 3

Leonor Cerdá Alberich, Valencia/ES
Clemens Cyran, Munich/DE
Anders Nordell, Stockholm/SE

 


17:05-17:15

Presentation 4 – SHARING
Technical and organizational obstacles and solutions for secure data platform in cancer imaging

Moderation: Emanuele Neri, Pisa/IT
Lauren Fromont, Barcelona/ES

 

17:15-17:55

Panel Session 4

Lauren Fromont, Barcelona/ES
Ignacio Gómez-Rico Junquero, Valencia/ES
Karim Lekadir, Barcelona/ES
Emanuele Neri, Pisa/IT
Gianna Tsakou, Athens/GR
Manolis Tsiknakis, Heraklion/GR

 


17:55-18:00

Closing

Emanuele Neri, Pisa/IT

EuCanImage

Enhancing the potential of Artificial Intelligence in cancer research

 

EuCanImage will build a highly secure, federated and large-scale European cancer imaging platform, with capabilities that will greatly improve capabilities of artificial intelligence (AI) in oncology. Firstly, the EuCanImage platform will be populated with a completely new data resource totaling over 25,000 single subjects, which will allow to investigate unmet clinical needs e.g., the detection of small liver lesions and metastases of colorectal cancer, or estimating molecular subtypes of breast tumours and pathological complete response. Secondly, the cancer imaging platform will be cross-linked to biological and health repositories through the European Genome-phenome Archive, allowing to develop multi-scale AI solutions that integrate organ-level, molecular and other clinical predictors into dense patient specific cancer fingerprints.

Facts & Figures, Consortium, Work Packages

Facts & Figures

Name: A European Cancer Image Platform Linked to Biological and Health Data for Next- Generation Artificial Intelligence and Precision Medicine in Oncology
Acronym: EuCanImage
Start Date: October1, 2020
End Date: September 30, 2024
Coordinator: Dr. Karim Lekadir (University of Barcelona)
Consortium: 20 partners from 11 countries
Funding: € 9.994.358,50

Work Packages

WP1: Legal and ethical framework for oncologic imaging

This WP will

  • Define the overall legal and ethical scope, issues and requirements of the EuCanImage platform to ensure a privacy-by-design approach to its development.
  • Establish a legal governancefor transnational and transcontinental health data transactions to guarantee the compliance of platforms operations across jurisdictions.
  • Establish internal and external policies to govern interactions between data owners, users and data processors in the platform.
  • Establish a novel ethical framework for developing and utilising AI-supported image-based decision support tools in clinical oncology.
  • Evaluate and deploy new cost-effective incentives for data owners to share cancer imaging data in exchange for value in the area of cancer care, while addressing societal, legal and economic concerns.

 

WP2: Clinical use cases, requirements and feedback

This WP will

  • Gather clinical requirements from the clinical partners, in collaboration with the data management and AI experts, on data management, AI development and AI assessment.
  • Use, evaluate and optimise the data deposition, curation and enhancement capabilities of the platform.
  • Design and implement AI solutions for the clinical use cases, and continuously refine them based on clinical feedback.
  • Assess the AI solutions from both technical and clinical perspectives.
  • Gather feedback from user experience, then provide recommendations and guidelines

 

WP3: Data platform and catalogue for cancer imaging and non-imaging data

This WP will

  • Define the overall legal and ethical scope, issues and requirements of the EuCanImage platform to ensure a privacy-by-design approach to its development.
  • Establish a legal framework for transnational and transcontinental health data transactions to guarantee the compliance of platforms operations across jurisdictions.
  • Establish internal and external policies to govern interactions between data owners, users and data processors in the platform.
  • Establish a novel ethical framework for developing and utilising AI-supported image-based decision support tools in clinical oncology.
  • Evaluate and deploy new cost-effective incentives for data owners to share cancer imaging data in exchange for value in the area of cancer care, while addressing societal, legal and economic concerns.

 

WP4: Suite for cancer imaging data curation, annotation and enhancement

This WP will

  • Create a comprehensive suite of open source tools and procedures for data anonymization that meet the legal requirements of all EU partners.
  • Integrate an existing cloud-based tool for collaborative and user-friendly annotation of cancer imaging data.
  • Expand POSDA tools and curation procedures to curate labelled data and non-imaging data.
  • Further enhance the wealth of the available data through synthetic image generation.
  • Implement tools for standardising image data cross sites and scanners.
  • Leverage the large-scale data and machine learning to provide semi-automated capabilities for data curation and annotation.

 

WP5: Artificial intelligence development platform and interfaces

This WP will

  • Provide the baseline compute environment to build flexible AI solutions for cancer imaging.
  • Build a comprehensive and scalable cancer radiomics library.
  • Implement a comprehensive machine learning toolbox for building integrative AI solutions.
  • Ensure the machine learning techniques can be executed in a distributed privacy-preserving manner.
  • Develop and validate tools for allowing the interpretability of AI-based decision.
  • Establish and document in detail the FUTURE Guiding Principles for AI in cancer imaging.

 

WP6: Open-access platform for assessing and benchmarking AI solutions in cancer imaging

This WP will

  • Achieve multi-stakeholder consensus with experts on metrics and criteria to assess and benchmark AI solutions.
  • Develop methods to estimate bias and uncertainty, as well as procedures to handle errors in AI for cancer imaging based on ensemble and online learning.
  • Develop new methods to assess the degree of interpretability of AI models.
  • Propose guidelines to assess clinical effectiveness and usability of AI solutions in clinical oncology.
  • Develop an open-access and disease-specific tool to evaluate cost-effectiveness of the AI products.
  • Integrate the assessment methods and procedures for community auditing and benchmarking within ELIXIR’s OpenEBench platform.

 

WP7: Project dissemination, communication & exploitation

This WP will

  • Develop EuCanImage’s corporate identity, as well as dissemination material.
  • Develop and iteratively update a plan for dissemination and communication.
  • Disseminate the project and its results to clinical, research and industrial stakeholders, as well as to the wider public.
  • Establish the network of image-based cancer researchers, clinicians and innovators.
  • Establish a plan for the exploitation and sustainability of the EuCanImage platform.

 

WP8: Scientific coordination and project management

This WP will

  • Monitor the successful implementation of research activities (WPs 1-7) within the agreed time, cost and quality limits, including management of risks and corrective actions.
  • Facilitate full synergy and interaction among EuCanImage partners by managing the internal channels, managerial committees, regular online and face-to-face meetings and project events.
  • Coordinate and manage all administrative, financial and contractual aspects related to the project.
  • Identify any upcoming risk as early as possible and work on mitigation plans and solutions.
  • Create synergies with other initiatives in the field and/or part of this call and in the field of cancer/imaging across Europe and the world.

 

WP9: Ethics requirements

This work package sets out the ethics requirements definded by the European Commision that the project must comply with.

Context

Currently, nearly all cancer treatments are guided based on human expertise and medical images. The clinical translation of existing AI-based cancer imaging solutions is still lacking, as all-too-often, they have been built and validated in small site-specific cancer imaging datasets. To increase trust, clinical value and translation of emerging AI technologies for cancer imaging, there is a need not only for datasets that are larger and more diverse, but also for new standards and best practices to leverage such large datasets for addressing unmet clinical needs in oncology. However, access to multi-centre cancer imaging dataset for the AI communities and industries remains a challenge as large cancer imaging repositories are lacking in Europe.

EuCanImage responds to the imperative need for high-quality and large repositories in Europe, and will additionally, take into account the European landscape in legal and ethical issues, variety of clinical systems, and existing data infrastructures.

To deliver this platform, we will build upon several key European initiatives in high-quality data sharing for personalised medicine research, including Euro-BioImaging and the European Genome-Phenome Archive. Furthermore, we’re working together with The Cancer Imaging Archive, a well-established cancer imaging repository in the US. This allows us to leverage their unique years-long experience in cancer imaging storage, curation and management. Our close collaboration between world-renowned experts in cancer research, AI and bioethics will establish necessary guidelines for developing standardised, trusted and transferable decisions support systems in future clinical oncology.

Research results

We’re making our research findings available free of charge for readers and are providing open access to published papers and reports. The list will be updated as the project progresses.

Development and external validation of a non-invasive molecular status predictor of chromosome 1p/19q co-deletion based on MRI radiomics analysis of Low Grade Glioma patients

Author(s): R. Casale et al.
Journal: European Journal of Radiology
Date: 5 April, 2021
DOI: https://www.ejradiology.com/article/S0720-048X(21)00158-3/fulltext

Data preparation for artificial intelligence in medical imaging: A comprehensive guide to open-access platforms and tools

Author(s): O. Diaz et al.
Journal: Physica Medica
Date: March, 2021
DOI: https://www.sciencedirect.com/science/article/pii/S1120179721000958#ak005

Prognostic and Predictive Value of Integrated Qualitative and Quantitative Magnetic Resonance Imaging Analysis in Glioblastoma

Author(s): M. Verduin et al.
Journal: Cancers
Date: 10 Februray, 2021
DOI: https://doi.org/10.3390/cancers13040722

Non-invasive imaging prediction of tumor hypoxia: A novel developed and externally validated CT and FDG-PET-based radiomic signatures

Author(s): S. Sanduleanu et al.
Journal: Radiotherapy and Oncology
Date: 1 November 2020
DOI: https://doi.org/10.1016/j.radonc.2020.10.016