Fighting bowel cancer with AI

Colon cancer is one of the most common and deadliest forms of cancer in Germany. Around 58,000 people are diagnosed with it every year. Recognised early, bowel cancer is easily curable. However, despite considerable progress in prevention and treatment, doctors still face challenges when it comes to diagnosis and prognosis. This is where the joint project DECADE – „Decentralised artificial intelligence for diagnosis, prognosis and therapy prediction in colorectal cancer“ comes in.

Multiple German university hospitals are jointly researching how the use of artificial intelligence (AI) and swarm learning (SL) can significantly improve the care and treatment of colorectal cancer patients in early and advanced stages. AI is already capable of analysing large amounts of data and recognising certain patterns. The insights gained from this can help to better predict the course of the disease or make more personalised diagnoses. The aim of this research project is to improve the treatment of colorectal cancer patients with the help of AI and SL.

Prof. Jakob N. Kather, project leader and Professor of Clinical Artificial Intelligence at the Else Kröner Fresenius Centre for Digital Health at TU Dresden and Dresden University Hospital, said at the start of the project: "AI tools have so far only been used sporadically in routine clinical practice. One reason is that the exchange of data between hospitals is severely restricted by legal and ethical barriers, particularly in Germany. One solution to this problem is swarm learning. With swarm learning, several institutions can train medical AI models together without exchanging data. By using decentralised artificial intelligence and swarm learning, we hope to improve the diagnosis, prognosis and treatment planning of colorectal cancer patients.“

Training AI with decentralised patient data

Data protection and ethical hurdles make it difficult to share sensitive patient data between different research institutions in cancer research, even though many patients are generally in favour of their data being used for research purposes. Swarm learning makes it easier to fulfil data protection requirements. Swarm learning is a special form of machine learning in which models are trained without the actual data having to be exchanged between the participants. The models are coordinated and merged via a blockchain, eliminating the need for a central instance. The DECADE project builds on this method to use SL-based AI technology to solve real-world clinical problems related to colorectal cancer.

„The legal requirements for the protection of sensitive health data are high. This innovative method of swarm learning makes it possible to utilise the benefits of collaboration and knowledge transfer between different research institutions without violating data protection regulations. In this way, AI models in cancer research can be further developed and improved to enable better diagnoses, prognoses and personalised treatment approaches for cancer patients,

said Prof. Tom Lüdde, Director of the Department of Gastroenterology, Hepatology and Infectiology at the University Hospital of Düsseldorf.

The project partners will use SL to develop AI algorithms for the diagnosis and subtyping of colorectal cancer and for predicting the course of the disease. In this way, they are creating a precedent for the use of SL in medicine that can serve as a template for any AI system in the healthcare sector. This is because more powerful AI systems can help doctors to detect colorectal cancer at an earlier stage and treat it more efficiently. This could support medical staff and improve the care and treatment of colorectal cancer patients.

Background information:
The research project DECADE – Decentralised Artificial Intelligence for Diagnosis, Prognosis and Therapy Prediction in Colorectal Cancer is being funded by German Cancer Aid with around 1.5 million euros over three years (2023-2026). The project partners are the university hospitals in Bonn, Dresden, Dusseldorf, Heidelberg and Mainz. The aim of the research project is to improve the treatment of colorectal cancer patients with the help of artificial intelligence and swarm learning.

Project partners:
• Else Kröner Fresenius Zentrum für Digitale Gesundheit, TU Dresden, Universitätsklinikum Carl Gustav Carus Dresden
• Universitätsklinikum Heidelberg, Pathologisches Institut, Abteilung für angewandte Tumorbiologie
• Universitätsmedizin Mainz, Institute of Pathology
• University Medical Centre Bonn, Medical Clinic and Polyclinic 1, National Centre for Hereditary Tumour Syndromes
• University Medical Centre Düsseldorf, Clinic for Gastroenterology, Hepatology and Infectiology

Article from the "innovations report" from 07/07/2023