Project Details
Projekt
Quantitative DWI und QSM zur Charakterisierung der Gewebemikrostruktur
Subject Area
Medical Physics, Biomedical Technology
Radiology
Radiology
Term
since 2023
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 500888779
This project is part of the research unit (RU) “Fast Mapping of Quantitative MR biosignatures at Ultra-high Magnetic Field”. It focuses on extending, accelerating, and improving diffusion and quantitative susceptibility magnetic resonance imaging. The work programme is split into two parts. In the first part, we prepare an accelerated protocol for the clinical projects of the RU. In the second part, further acceleration and quality improvements shall be achieved. Specifically, we will implement a local low rank regularized echo planar imaging sequence for diffusion-weighted imaging. It uses a patch-wise regularization and exploits data redundancies in acquisitions with multiple diffusion encodings to effectively increase the signal-to-noise ratio and therewith speed up the acquisition process. We will develop a sequence that features essentially arbitrary diffusion encoding capabilities (such as b-tensor encoding). In the second step, we will develop an interleaved multi-shot version of this sequence to reduce image distortions that are prevalent in 7 Tesla echo planar imaging; and to speed-up the acquisition by means of shorter readout trains. For quantitative susceptibility mapping (QSM), we will implement a sequence with a stack-of-stars sampling trajectory. Since the magnitude images of gradient echo sequences acquired at different echo times exhibit data redundancies comparable to those of diffusion-encoded images, we will again use a local low rank regularization in the image reconstruction. The radial trajectories of this sequence should be well-suited for an undersampled and therewith accelerated image reconstruction. In a second step, we will enhance the capabilities of our sequence with a quasi-continuous echo time sampling, where each spoke has its own optimized echo time. This will allow an increased quality of QSM when fat is present in the image, as it is often the case in muscle exams and breast imaging. Moreover, the sequence will be able to yield breathing cycle resolved field maps. On the QSM reconstruction side, we will develop deep learning-based approaches to enable a high-quality reconstructions with a reduced amount of image data compared to conventional reconstruction approaches. We will adapt existing neural networks from lower field strengths to 7 T and enhance their capabilities so that we can also integrate breathing cycle dependent field maps and quasi-continuous echo times. This project will receive parallel transmission methods (pTx) from the pTx project of the RU. We will deliver the developed sequences to the RU’s clinical projects after the first year. Moreover, we will forward essential evaluation and image reconstruction methods to the other projects of the RU.
DFG Programme
Research Units
