Published on 13.03.2023

AI Award for Helmholtz Imaging and DKFZ researcher Tim Rädsch

Portrait of Tim Rädsch, part of the Research Unit at DKFZ

The Technical University of Deggendorf announced the Anton Fink-Wissenschaftspreis für Künstliche Intelligenz (Anton Fink Science Award for Artificial Intelligence) for the second time in 2022.

Now the prize winner has been determined: Tim Rädsch from the German Cancer Research Center and Helmholtz Imaging convinced the jury the most. He will now receive the prize money of 10,000 euros for his research.

Out of 18 AI research teams from Germany and Austria Tim Rädsch was selected as winner. The final thesis of his master’s degree at the Karlsruhe Institute of Technology was written in the DKFZ Department of Intelligent Medical Systems. Rädsch presented the first systematic study on labeling instructions in biomedical imaging.

AI algorithms learn from images in which relevant structures (e.g., tumors) are marked – experts refer to this as annotations. A labeling instruction is a document that specifies the correct marking of relevant structures in biomedical images. Raedsch was able to demonstrate that the type of instruction provided is critical to the quality of datasets that form the foundation for AI algorithms. This is in stark contrast to the common practice of annotating without instructions in a large proportion of biomedical studies. A scientific paper of his findings has just been accepted in the journal Nature Machine Intelligence, underpinning the high quality of the work.

The award ceremony will take place on 11 May 2023.

Read the complete press release by DKFZ (in German)

More information about the THD AI Prize

Tim Rädsch is part of the Helmholtz Imaging Research Unit at DKFZ that focuses on the various challenges that currently impede advanced AI technologies from directly benefiting society. The unit pioneers research in three focus areas to address these challenges: advance to a deeper understanding and generalizability of algorithms; consider humans as an integral part of AI-application to enable reliable and safe deployment; and lead initiatives towards standardizing evaluation and benchmarking practices in the field and develop user-friendly tools to guide the community. More