RadioVal Work Packages

Aims and specific objectives

RadioVal develops a validated AI-based decision-making support system, increasing clinicians’ and patients’ trust in artificial intelligence tools by implementing the first international, clinical validation study of radiomics-based prediction of neoadjuvant chemotherapy (NAC) treatment response from breast MRI. This reduces overtreatment in patients undergoing chemotherapy and reduce costs of breast cancer care. To test applicability as well as transferability, the validation will take place in eight clinical centres from three high-income EU countries (Sweden, Austria, Spain), two emerging EU countries (Poland, Croatia), and three countries from South America (Argentina), North Africa (Egypt) and Eurasia (Turkey).

Work Package description

  • Develop a social innovation framework to compile multi-stakeholder requirements and pathways for enhancing radiomics evaluation and implementation in breast cancer.
  • Develop a radiomics information and communication package that will be leveraged during social innovation and radiomics evaluation by RadioVal’s multi-disciplinary stakeholders.
  • Leverage the social innovation framework to define multi-disciplinary requirements and pathways in the following specific areas:
    • clinical, patient and healthcare,
    • socio-ethical,
    • legal and regulatory.
  • Identify criteria and metrics for multi-faceted evaluation of radiomics AI, in breast cancer in particular.
  • Define stages, timelines and procedures for radiomics evaluation, incl. in-silico and external validations.
  • Establish an open-access reference dataset for comparative evaluation and community benchmarking of radiomics AI in the field of breast NAC.
  • Introduce a cost-effectiveness analysis approach that will model the specificities of radiomics AI tools.
  • Develop and disseminate new guidelines for the reporting of radiomics evaluation studies.
  • Develop the radiomics passport for standardised description and traceability of radiomics models.
  • Implement methods for both image and segmentation automated quality controls.
  • Develop methods for automated failure detection, especially out-of-distribution abnormal deviations.
  • Implement human-in-the-loop mechanisms to enable a degree of human oversight and to integrate clinical feedback to radiomics AI over the model’s lifetime.
  • Integrate all these methods into a ready-to-deploy radiomics traceability suite.
  • Perform a preliminary evaluation of ; the very first toolbox for radiomics monitoring and continuous evaluation.
  • Evaluate mitigation measures and suitable strategies to enhance the radiomics fairness, robustness, usability and explainability for treatment assessment in breast cancer.
  • Integrate the radiomics tools and AI optimisation methods into a functional Clinical Decision Support System (CDSS) that will be ready for clinical deployment and validation.
  • Produce the second version of the information and communication package integrating new information and guidelines on the tool’s clinical usability, as well as on radiomics fairness, robustness and explainability.
  • Implement an international study for in-depth, multi-faceted evaluation of the radiomics tools for treatment planning in breast cancer.
  • Evaluate and analyse in particular the robustness, fairness, usability, explainability, traceability and scalability of the radiomics models.
  • Compare the results between internal and external sites, as well as across international countries.
  • Gather feedback and report the results of the study according to the reporting guidelines.
  • Perform an in-depth cost-effectiveness analysis of the radiomics tool for breast cancer treatment and care.
  • Evaluate impact of the introduction of radiomics based treatment planning in breast cancer practice.
  • Evaluate the socio-ethical implications of the introduction of the radiomics tool in the real world.
  • Establish pathways for regulatory certification of radiomics AI tools for clinical translation.
  • Deliver the final version of the RadioPack information package by integrating clinical feedback.
  • Examine possible exploitation routes and sustainability plans to maximise impact beyond the project.
  • Manage the timely implementation of the research activities, administrative and financial tasks based on the frequently updated Data Management Plan (DMP).
  • Continuously disseminate and communicate the project and its results to clinical, research and industrial stakeholders, as well as to the wider public.
  • Organise scientific workshops and other outreach events throughout the duration of the project.
  • Create synergies with other initiatives in the field of radiomics, AI and/or breast cancer.