PRIMAGE project: predictive in silico multiscale analytics to support childhood cancer personalised evaluation empowered by imaging biomarkers

Luis Martí‐Bonmatí, Ángel Alberich‐Bayarri, Ruth Ladenstein, Ignácio Blanquer, J. Damián Segrelles, Leonor Cerdá-Alberich, Polyxeni Gkontra, Barbara Hero, José Manuel García‐Aznar, Daniel A. Keim, Wolfgang Jentner, Karine Seymour, Ana Jiménez-Pastor, Ismael González‐Valverde, Blanca Martínez de las Heras, Samira Essiaf, Dawn Walker, Michel Rochette, Marian Bubak, Jordi Mestres, Marco Viceconti, Gracia Martí-Besa, Adela Cañete, Paul Richmond, Kenneth Y. Wertheim, Tomasz Gubała, Marek Kasztelnik, Jan Meizner, Piotr Nowakowski, Salvador Gilpérez, A. Suárez-Garcı́a, Mario Aznar, Giuliana Restante, Emanuele NeriEuropean 2020 · DOI

Abstract

PRIMAGE is one of the largest research projects focusing on medical imaging, artificial intelligence (AI), and pediatric cancer treatment. This four-year European Commission-funded initiative involves 16 partners across Europe, including the European Society for Paediatric Oncology and major pediatric oncology units. The project establishes an observational in silico study using high-quality anonymized datasets—including imaging, clinical, molecular, and genetic data—to train and validate machine learning and multiscale algorithms. The resulting open cloud-based platform provides clinical assistance for diagnosis (phenotyping), treatment prediction, and prognosis. Initially validated on neuroblastoma and diffuse intrinsic pontine glioma, the platform is designed for future translation to other malignant solid tumors.

Introduction: Addressing the Data Gap in Pediatric Oncology

Pediatric cancers, particularly Neuroblastoma (NB) and Diffuse Intrinsic Pontine Glioma (DIPG), present significant challenges for clinical research. Due to their rarity, gathering sufficient data for robust clinical trials is difficult. While "big data" exists in the form of high-resolution imaging and genomic profiles, these datasets are often siloed and underutilized.

Clinicians currently lack integrated tools to synthesize these heterogeneous data types into precise, individualized predictions. There is a critical need for decision support systems (DSS) that can help determine whether a tumor will regress spontaneously or require aggressive intervention. PRIMAGE addresses this by creating a computational framework to bridge the gap between complex data and clinical action.

The PRIMAGE Technological Framework

The project is built upon a hybrid Cloud/High-Performance Computing (HPC) platform designed to handle massive datasets and complex simulations.

The PRIMAGE platform concept diagram showing the integration of cloud infrastructure and clinical decision support.

The architecture includes:

  • PRIMAGE Core: Houses the simulation frameworks and models for diagnosis and tumor growth.
  • Middleware: Connects the core to European clinical registries, biobanks, and cloud infrastructures like EOSC and EuroHPC.
  • Web Platform: A user-centric interface for data upload, patient browsing, and visualization of results with confidence scores.

Multiscale Modeling and AI-Driven Insights

At the heart of PRIMAGE is a multiscale approach that links biological processes across three levels:

  1. Cell Scale: Modeling chemical and biological properties within a single cancer cell.
  2. Tissue Scale: Using agent-based models to track cell interactions and population dynamics.
  3. Tumor Scale: Simulating volume evolution and biomechanical interactions with surrounding tissues.
The PRIMAGE pipeline from image acquisition to improved health outcomes.

By treating medical images as "virtual biopsies," the platform uses radiomics and deep learning to extract quantitative biomarkers. These features are correlated with genomic profiles to provide a holistic, non-invasive view of the tumor.

Datasets and Clinical Validation

The project leverages extensive anonymized datasets from 16 European partners:

  • Neuroblastoma (NB): Approximately 1,000 cases from the GPOH (German Society of Paediatric Oncology and Haematology) database.
  • DIPG: Approximately 700 cases from the SIOPE (European Society for Paediatric Oncology) registry.
  • Integrated Dataset: A target sample of ~900 cases featuring a complete mix of imaging (MRI, CT, PET), histology, and longitudinal clinical data.

The consortium has prioritized five critical clinical endpoints (CEPs) for validation, such as predicting spontaneous regression in low-risk NB and estimating survival periods for DIPG patients.

Conclusion and Future Directions

PRIMAGE represents a shift toward open science in oncology. By utilizing the European Open Science Cloud (EOSC), the project aims to provide a global resource for the scientific community. While data heterogeneity and the need for rigorous external validation remain challenges, the successful implementation of this platform for NB and DIPG will pave the way for its translation to other malignant solid tumors in both children and adults.

Summary updated

Abstract

PRIMAGE is one of the largest and more ambitious research projects dealing with medical imaging, artificial intelligence and cancer treatment in children. It is a 4-year European Commission-financed project that has 16 European partners in the consortium, including the European Society for Paediatric Oncology, two imaging biobanks, and three prominent European paediatric oncology units. The project is constructed as an observational in silico study involving high-quality anonymised datasets (imaging, clinical, molecular, and genetics) for the training and validation of machine learning and multiscale algorithms. The open cloud-based platform will offer precise clinical assistance for phenotyping (diagnosis), treatment allocation (prediction), and patient endpoints (prognosis), based on the use of imaging biomarkers, tumour growth simulation, advanced visualisation of confidence scores, and machine-learning approaches. The decision support prototype will be constructed and validated on two paediatric cancers: neuroblastoma and diffuse intrinsic pontine glioma. External validation will be performed on data recruited from independent collaborative centres. Final results will be available for the scientific community at the end of the project, and ready for translation to other malignant solid tumours.
BibTeX
@inproceedings{martibonmati2020primage,
  author = {Luis Martí‐Bonmatí and Ángel Alberich‐Bayarri and Ruth Ladenstein and Ignácio Blanquer and J. Damián Segrelles and Leonor Cerdá-Alberich and Polyxeni Gkontra and Barbara Hero and José Manuel García‐Aznar and Daniel A. Keim and Wolfgang Jentner and Karine Seymour and Ana Jiménez-Pastor and Ismael González‐Valverde and Blanca Martínez de las Heras and Samira Essiaf and Dawn Walker and Michel Rochette and Marian Bubak and Jordi Mestres and Marco Viceconti and Gracia Martí-Besa and Adela Cañete and Paul Richmond and Kenneth Y. Wertheim and Tomasz Gubała and Marek Kasztelnik and Jan Meizner and Piotr Nowakowski and Salvador Gilpérez and A. Suárez-Garcı́a and Mario Aznar and Giuliana Restante and Emanuele Neri},
  title = {{PRIMAGE project: predictive in silico multiscale analytics to support childhood cancer personalised evaluation empowered by imaging biomarkers}},
  booktitle = {European},
  year = {2020},
  doi = {10.1186/s41747-020-00150-9},
}
Figure