DFG Funds New Research Project on the Theory of Learning and Motion-Compensated Image Reconstruction

The German Research Foundation (DFG) has approved funding for a new research project at the Pattern Recognition Lab (Chair of Computer Science 5) of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). Led by Professor Dr.-Ing. Andreas Maier, the project will investigate the theoretical foundations of learning with known operators and explore their applications in motion-compensated image reconstruction.

The primary goal of the project is to advance our theoretical understanding of hybrid learning approaches that incorporate known mathematical operators—such as physical models or system equations—into machine learning algorithms. These methods are especially relevant in medical imaging, where prior knowledge can be integrated to improve reconstruction quality and reduce motion artifacts, for instance in CT or MRI scans.

The DFG will support the project with a total funding amount of €271,051 plus a 22% program allowance over a duration of 36 months.

“This grant enables us to explore the fundamental limits of combining model-based and data-driven approaches. The findings are expected to provide both theoretical insights and practical advancements, particularly in the field of medical image reconstruction,”
– Prof. Dr.-Ing. Andreas Maier

The Pattern Recognition Lab gratefully acknowledges the DFG for supporting this forward-looking research endeavor.