Automatic, robust quantification of lung perfusion in DCE-MRI in a multicenter study with COPD patients (QuantLuPe)

• Applicants: Professor Dr. Hans-Ulrich Kauczor, since 2/2025; Dr.-Ing. Michael Selzer
• Subject Area: Medical Informatics and Medical Bioinformatics ; Nuclear Medicine, Radiotherapy, Radiobiology ; Radiology
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 554598213

Widespread clinical use of image-based perfusion measurements in lung diagnostics is hindered by the lack of a robust, automated perfusion quantification tool and by a shortage of radiologists who can evaluate contrast-enhanced MRI (DCE-MRI) to provide a qualitative assessment of pulmonary perfusion. The existing software solutions for perfusion quantification have so far failed to automatically evaluate MRI images of different acquisition conditions, such as different MRI devices or protocols. This gap should be closed by the development of QuantLuPe by further developing an existing perfusion quantification tool for analyzing a multicenter study with 600 patients with chronic obstructive pulmonary disease (COPD). To simplify the further development and application of QuantLuPe, the open-source research data management (RDM) and workflow tool Kadi4Mat is adapted to the requirements in radiological research in parallel and then introduced as Kadi4Rad. The research objectives of this study include a) the development of robust lung segmentation in multicenter studies, b) the development of a method for breathing motion correction in DCE-MRI images of the lung, c) the establishment and further development of a user-friendly research data management and workflow tool for sustainable radiological research and d) the further development of a tool for robust, automatic quantification of perfusion in DCE-MRI in a multicenter COPD study. First, alternative segmentation methods for the segmentation of the lung in the morphological images of the COPD cohort are developed and compared. In addition, a super-resolution reconstruction model is developed to improve the image quality before segmentation. Since measurements with breathing motion lead to problems in perfusion quantification, an algorithm for breathing motion correction in the DCE-MRI images is developed. For this purpose, two independent methods based on perfusion physics and motion tracking are developed and compared with each other. To simplify the usage of the developed methods, the existing RDM and workflow tool Kadi4Mat is optimized for radiological research. For this purpose, modules for the integration of image data are added and interfaces to existing radiological software tools are created within workflows. Finally, the optimized perfusion quantification tool is to be used for the evaluation of a multicenter COPD study.

• DFG Programme: Research Grants
• International Connection: USA
• Co-Investigators: Dr.-Ing. Arnd Koeppe; Professorin Dr. Britta Nestler; Dr. Oyunbileg von Stackelberg; Dr. Simon Triphan
• Cooperation Partner: Professor Craig Galban, Ph.D.
• Ehemaliger Antragsteller: Professor Dr. Mark Oliver Wielpütz, until 2/2025