Using Artificial Intelligence to Decipher the Mechanisms of Cancer Metastasis

From March 1, 2025, leading experts from the fields of medicine, computer science and biotechnology will work together in the DECIPHER-M project (Deciphering Metastasis with Multimodal Artificial Intelligence Foundation Models). Led by Prof. Dr. Jakob N. Kather at the Else Kröner Fresenius Center for Digital Health, the interdisciplinary research team is using AI to study the development and spread of cancer metastases. The AI models link complex data from different sources to enable precise diagnoses and more individualized treatment options. “Despite enormous progress in oncology, metastasis remains one of the biggest challenges in cancer treatment. In the DECIPHER-M project, we are using AI technologies to identify complex patterns in routine clinical data,” says Prof. Kather, project coordinator. “Our multimodal approach allows us to predict the individual risk of metastasis more precisely and to develop personalized treatment options. The long-term goal is to improve the survival of cancer patients.”

AI Systems Process Different Types of Data and Recognize Patterns  

The development of cancer metastases depends on multiple factors that are often difficult to identify. DECIPHER-M therefore relies on an AI model that combines different medical data sources – including tissue samples, X-ray and MRI images, and genetic information. These so-called multimodal foundation models link different types of data and recognize patterns. The insights gained help identify the risk of metastasis at an early stage and develop personalized treatment options. This enables more precise diagnoses, the initiation of preventive measures and the optimization of treatment options for cancer patients.

Julia Schnabel, Director of the Institute of Machine Learning in Biomedical Imaging at Helmholtz Munich, plays a key role in this effort. Her team is developing a Cancer Foundation Model that integrates pathology, radiology, text reports, and electronic health records. By combining Vision Transformers (AI models specialized in image analysis) with Large Language Models (LLMs), their systems will analyze imaging and text data to accurately identify the origin of cancer.

BMBF Funding for Pioneering Interdisciplinary Research

“For Dresden Medical Faculty, interdisciplinary collaboration is the key to success in research and patient care. The project partners from Aachen, Dresden, Essen, Heidelberg, Mainz and Munich are collaborating to improve the quality of cancer treatment, avoid ineffective treatments and reduce the burden on the health care system. In the long term, DECIPHER-M could help to reduce cancer mortality and significantly improve the quality of life of cancer patients,” said Prof. Dr. med. Dr. Esther Troost, Dean of the Carl Gustav Carus Faculty of Medicine at the TU Dresden. DECIPHER-M has been submitted for a five-year period (2025–2030) with a total budget of around €9 million. The German Federal Ministry of Education and Research (BMBF) will initially fund the project for three years with around €5.5 million. Helmholtz Munich will be funded with approximately €544,000 for this first funding period.