Quantitative metabolic MR biomarker imaging at 7T using end-to-end bio-physics-informed optimization and inference

• Applicant: Professor Dr. Moritz Zaiss
• Subject Area: Medical Physics, Biomedical Technology
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 559955894

Alterations of intracellular pH and protein, lipid, and metabolite concentrations underlie a variety of diseases, including cancer. Importantly, pathology-related effects at the molecular-level are manifested long before any structural changes occur and may also reflect the host-response to therapeutics. The currently employed in-vivo molecular imaging techniques, however, either require the use of a radioactive/metal-based contrast-agent, a lengthy acquisition protocol, or provide limited sensitivity/resolution. Since gaining regulatory clearance in 2017, 7T MRI opened new opportunities for multi-metabolite molecular imaging based on the chemical exchange saturation transfer (CEST) mechanism. At 7T, the signal-to-noise ratio and spectral selectivity are drastically improved, allowing better separation of signal sources and detection of metabolites invisible at lower fields. However, harnessing high-field MRI for CEST imaging presents unique challenges, including increased radiofrequency energy deposition and field inhomogeneity, which leads to artifacts in the generated images, and limits accurate modelling of the signals. Moreover, the contrast-weighted nature of traditional CEST imaging, its long acquisition time, and associated readout effects all acted as major barriers to the widespread adoption of CEST MRI for practical human studies. The central goal of this project is to develop a transformative and autonomous ultra high-field CEST-MRI-based technology, for specific, quantitative, and rapid multi-metabolite cancer imaging. Our working hypothesis is that synergistically integrating biophysical models (which explain the effects of various metabolites and proteins on the molecular 7T MRI signal) and imaging process models, with a novel machine learning-based optimization technique throughout the entire contrast preparation and image encoding pipeline, will enable accurate characterization of tissue state. We will leverage three biophysics-driven AI disruptive technologies (MRzero, AutoCEST, and CEST MR-Fingerprinting) recently introduced at Germany (Zaiss Lab, Erlangen University) and the Middle East (Perlman Lab, Tel Aviv University, Israel), alongside the complimentary expertise of our two labs, to form a strong foundation for the current project. First, we will establish an end-to-end method for the discovery of rapid 3D multi-metabolite 7T imaging protocols and parameter quantification. Second, we will validate the rigor and reproducibility of the method, in a multi-site test-retest study. Third, the method will be used to determine the pH and multi-metabolite dynamics in brain cancer patients. The optimized imaging technique will constitute a valuable tool for shedding new light on the molecular mechanisms underlying brain cancer, and will ultimately be expandable for a variety of additional in-vivo molecular imaging tasks.

• DFG Programme: Research Grants
• International Connection: Israel
• International Co-Applicant: Or Perlman, Ph.D.