Revolutionising paediatric cancer care through AI-driven imaging biomarkers
Digital transformation in healthcare leverages technological advances to enhance patient care, streamline administrative processes, and improve overall efficiency of patient-centred healthcare services. This transformation includes the adoption of electronic health records, telemedicine solutions, AI-driven diagnostics, and data analytics. Alongside advances in medical imaging, they can improve the diagnosis and treatment of cancer, enhancing quality of life.
A cloud platform for personalised disease management
The EU-funded PRIMAGE project was designed to assist diagnosis, prognosis and therapy in children with neuroblastoma and diffuse intrinsic pontine glioma (DIPG), an aggressive tumour located at the base of the brain. The key objective was to develop a cloud-based, high-performance computing platform that assists personalised clinical management. Researchers employed retrospective data (imaging, clinical, molecular and genetics) from various European paediatric oncology units and the European Society for Paediatric Oncology, to build AI models and integrate them into a comprehensive decision support tool. “Our main goal was to translate multidisciplinary and multiscale data into predictors for personalised decision making,” states project coordinator Luis Martí-Bonmatí.
Computational analysis of medical images
Oncologic imaging is an ideal modality for the exploration and validation of novel biomarkers, given its frequent use in cancer depiction, classification, staging and monitoring of treatment response. The consortium extracted quantitative imaging and molecular biomarkers from tissue and liquid biopsies from patients with neuroblastoma and DIPG and reused them in imaging biobanks. The project created a common framework for collecting and assessing these biomarkers, overcoming technical challenges associated with obtaining reproducible imaging data in a standardised format to extract generalisable trends. They successfully collected data from 1 148 cases of neuroblastoma and 71 cases of DIPG and identified radiomics features that correlated with specific clinical endpoints. “Neuroblastoma and DIPG are rare paediatric tumours, so it was quite challenging to collect a sufficient amount of data to allow AI model training,” admits Martí-Bonmatí. This process ultimately led to the development and training of AI models that integrate molecular, biological, and genomic biomarkers with imaging and clinical data to predict various clinical outcomes. Additionally, they developed multiscale models for simulating tumour growth and integrated an advanced visualisation environment in the infrastructure. “PRIMAGE is the first project that developed protocols for image processing and successfully identified image biomarkers to build a decision support system for these paediatric patients with cancer,” emphasises Martí-Bonmatí.
Joint European ventures for cancer imaging
PRIMAGE has joined other EU-funded projects (Chaimeleon, EuCanImage, INCISIVE, ProCancer-I) in an Artificial Intelligence for Health Imaging (AI4HI) network that spans 20 countries and aims to establish AI solutions for medical imaging analysis and interpretation. Combined with major European research infrastructures, AI4HI aims to integrate real world data and implement robust and ethically and legally compliant AI solutions for cancer diagnosis. The next step for the PRIMAGE platform is to be integrated into the European Federation for Cancer Images (EUCAIM), the largest EU cancer imaging infrastructure created to date, aiming to achieve more precise and faster clinical decision making. “Through this integration, we expect the PRIMAGE project partners to be able to continue collecting data and doing collaborative research for years to come,” concludes Martí-Bonmatí.
Keywords
PRIMAGE, AI, diagnosis, neuroblastoma, diffuse intrinsic pontine glioma (DIPG), cloud platform, imaging, biomarker
