Quantitative T2* Relaxation and Field Inhomogeneity Mapping Using Multi-Echo Radial FLASH Magnetic Resonance Imaging and Model-based Iterative Reconstruction

• Applicants: Dr. Zhengguo Tan; Professor Dr. Martin Uecker
• Subject Area: Medical Physics, Biomedical Technology ; Nuclear Medicine, Radiotherapy, Radiobiology
• Term: from 2019 to 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 427934942

• Final Report Year: 2023

Final Report Abstract

In this project, we developed a new technique for magnetic resonance imaging (MRI) that acquires quantitative information related to fat and R2* (related to iron content) in the human body. This technique combines a new fast radial scanning technique that acquires data continuously during free breathing with new algorithms for image reconstruction. Data is first sorted into respiratory bins using a previously developed self-gating method. As a reconstruction of the time-resolved maps of water, fat, the R2* relaxation is then performed directly from the acquired data without additional post-processing steps by formulating the reconstruction as a high-dimensional non-linear inverse problem. This has two advantages: First, by modeling the physics of the measurement process, multiple echoes can be used which improves scan efficiency. Second, because the mathematical framework allows the integration of prior knowledge into the reconstruction by choosing regularization terms, less data is required further reducing scan time. We implemented different physical models into the open-source toolbox BART. Next, we built water, fat, and iron phantoms and validated the techniques against reference measurements. We tested the method with in vivo data for different applications, looking at the iron quantification of the human heart, liver fat quantification, and imaging of brain function. We found that good initialization of the numerical algorithm and non-linear optimization of the magnetic B0 field map is required to obtain accurate fat-fraction and R2∗ maps. Nevertheless, inhomogeneities of the B0 field still pose a challenging problem for some application (e.g. cardiac imaging), a problem that could not be completely solved in this project. We then focused on free-breathing liver imaging and studied the use of different regularization techniques to improve image quality and reduce scan time. Here, we found that the use of advanced spatio-temporal regularization terms could improve image quality. Finally, in a small study in volunteers with suspected non-alcoholic fatty liver disease, we could show that the new free-breathing acquisition with advanced reconstruction is more robust than standard breathhold scans for quantification of liver fat.

Publications

• Free- Breathing Liver Fat and R2* Mapping: Multi-Echo Radial FLASH and Model-based Reconstruction (MERLOT). Proceedings of International Society for Magnetic Resonance in Medicine 2021; 30:0753.
  Tan Z., Rosenzweig S., Wang X., Scholand N., Holme H.C.M., Blumenthal M. & Uecker M.
  
• Physics-based reconstruction methods for magnetic resonance imaging. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 379, 20200196.
  Wang, Xiaoqing; Tan, Zhengguo; Scholand, Nick; Roeloffs, Volkert & Uecker, Martin
  
• Radial Echo-Planar Imaging with Subspace Reconstruction for Brain MRI. ISMRM 31st Annual Meeting. ISMRM.
  Tan, Zhengguo & Uecker, Martin
  
• Free-Breathing Liver Fat,R₂* andB₀ Field Mapping Using Multi-Echo Radial FLASH and Regularized Model-Based Reconstruction. IEEE Transactions on Medical Imaging, 42(5), 1374-1387.
  Tan, Zhengguo; Unterberg-Buchwald, Christina; Blumenthal, Moritz; Scholand, Nick; Schaten, Philip; Holme, Christian; Wang, Xiaoqing; Raddatz, Dirk & Uecker, Martin