CancerScan develops a smart digital pathology slide scanner that supports diagnosis and patient-specific treatment recommendations in oncology. By combining multi-omics profiling, clinical data, and computational modelling, the project aims to create tumour digital twins embedded directly within scanning hardware, enabling the simulation of drug responses and supporting more informed, individualised treatment decisions.

CancerScan is a Horizon Europe research project addressing the challenge of predicting how cancers respond to treatment. Tumour behaviour and drug effectiveness are strongly influenced by the surrounding tumour microenvironment, the mix of cells and signals around a tumour, yet this complexity is difficult to capture with current diagnostic tools. The project focuses on turning detailed information from pathology slides into practical support for clinical decision-making that can be used during routine diagnostic workflows.

The core innovation of CancerScan is the development of a smart pathology slide scanner that can semi-automatically generate patient-specific tumour digital twins. These digital twins combine detailed biological information from biopsy slides with standard clinical data, allowing different drug treatments to be tested on a virtual model of the tumour before a therapy is chosen in the clinic. The digital twin functionality is built directly into the scanner hardware to enable efficient analysis and simulation without adding extra steps for pathology laboratories.

A key scientific focus of the project is the study of tumour communication. CancerScan examines the interactions between cells within the tumour and its surrounding environment, and how these interactions affect how well drugs work, helping clinicians understand why treatments may succeed or fail in individual patients. Experimental findings are organised using biomedical knowledge graphs, which can be understood as structured maps linking biological processes, treatments, and outcomes, bringing together different types of data into a clear and structured representation of tumour behaviour and treatment response.

To simulate treatment effects, the project develops mathematical and computational models that describe how tumour communication changes over time, allowing researchers to explore how tumours might evolve under therapy. Statistical analysis and machine learning are used to identify important patterns that influence treatment outcomes, supporting evidence-based treatment choices. Pancreatic cancer is used as a model case, combining laboratory-grown tumour models with patient biopsy data to ensure relevance for real diagnostic situations.

CancerScan represents an initial step towards a platform that supports healthcare professionals in diagnosis and in evaluating the effectiveness of drug treatments for individual patients, with the potential to reduce trial-and-error treatment decisions, and with the long-term goal of improving personalised cancer care and patient outcomes.

Project objectives

To achieve this vision, CancerScan brings together experimental research, data integration, and advanced modelling within a single diagnostic platform. The project pursues the following objectives:

• Map how the tumour microenvironment influences the effectiveness of cancer drug treatments.
• Develop structured knowledge models that combine experimental and clinical data to describe tumour behaviour and treatment response.
• Identify key patterns in tumour communication using statistical analysis and machine learning.
• Design and validate a platform for the automated creation of tumour digital twins for specific tumour situations.
• Create an embedded hardware–software system that enables simulation of treatment outcomes from digital pathology slides.
• Increase awareness of project outcomes through communication and dissemination activities.

Consortium

CancerScan is a 36-month project involving 8 institutions across 6 countries. Coordinated by Agencia Estatal Consejo Superior de Investigaciones Científicas from Spain, the project includes partners from Portugal, Italy, Germany, Austria, and Serbia.

Links

https://www.cancerscanproject.eu/

http://www.linkedin.com/company/cancerscan-project

https://x.com/CancerScan_eu

https://www.youtube.com/@cancerscan-Project

Keywords

oncology, digital pathology, tumour digital twins, personalised treatment, slide scanner, tumour microenvironment, drug response simulation, machine learning