Artificial Intelligence to optimize the Analysis of MRI Images
- News
- Research
![[Translate to English:]](/storages/physik/_processed_/1/e/csm_KI-MRT_d0f22f224e.jpeg "[Translate to English:]")
Scientists from the Department of Medicine at the University of Duisburg-Essen (UDE) and physicists from TU Dortmund University have launched the joint research project "k-Radiomics". The team is developing new methods of artificial intelligence and machine learning to be able to evaluate MRI images even better in future and thus better characterize tumours and tissue. The project is being funded by the Bruno and Helene Jöster Foundation with around 2.6 million euros until 2028.
In the project, the interdisciplinary research team is focusing on the raw data from magnetic resonance imaging (MRI) scans, in particular the so-called k-space. Until now, this data has not been sufficiently interpretable and therefore cannot be fully used in clinical practice to characterize tumours and tissues - and ultimately to improve the diagnosis and treatment of diseases. The team from Essen and Dortmund is therefore developing new AI methods for the use of such raw data. The ultimate aim is to enable improved tissue characterization in the sense of "virtual biopsies".
Prof. Jens Kleesiek, Prof. Jan Egger and Moritz Rempe from the Institute for Artificial Intelligence in Medicine at Essen University Hospital are leading the project at UDE; Prof. Kevin Kröninger and his team from the Faculty of Physics at TU Dortmund University are involved. The Dortmund working group is developing special machine learning methods in the project: so-called generative neural networks that can process the complex raw data from the MRIs. These methods are then used to characterize tissue and compared with traditional methods. At the end of the project, the methods developed will be made available to other researchers as open source libraries.
Contact for further questions:
