Schwarz, Jolanda M.: Advanced Image Reconstruction Methods for Ultra-High Field MRI. - Bonn, 2020. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.
Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5-58632
@phdthesis{handle:20.500.11811/8442,
urn: https://nbn-resolving.org/urn:nbn:de:hbz:5-58632,
author = {{Jolanda M. Schwarz}},
title = {Advanced Image Reconstruction Methods for Ultra-High Field MRI},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2020,
month = jul,

note = {Magnetic resonance imaging (MRI) has become a very powerful and flexible medical imaging technique, that allows an unrivaled insight into human anatomy and physiology. For a wide range of possible applications, acquisition of high-quality images in short scan times is essential. A recent step in this direction is the introduction of the first clinical 7 Tesla MRI scanners. For instance, the high signal-to-noise ratio (SNR) achieved by the strong magnetic field allows for increased parallel imaging accelerations. However, the images are more susceptible to magnetic field deviations, such as those caused by subject motion or instrumental imperfections. This thesis addresses these challenges and contributes to the acquisition and reconstruction of ultra-fast high-quality 3D images at 7 Tesla MRI scanners.
The first part focuses on the development and validation of a novel non-iterative parallel imaging reconstruction for wave-CAIPI acquisitions. The recently proposed wave-CAIPI sampling strategy along corkscrew sampling trajectories in the spatial frequency space allows to utilize the coil sensitivity variations of multiple receiver coils in all three dimensions, instead of two with conventional parallel imaging. Although Cartesian parallel imaging reconstructions are no longer applicable, the reconstruction can be formulated as a Cartesian problem which allows to use a GRAPPA-based reconstruction of the missing data in the frequency space. The developed GRAPPA-based wave-CAIPI reconstruction is fast and robust and, compared to the previously proposed iterative SENSE-type reconstruction, it does not depend on the accuracy of specific coil sensitivity estimations and mask regions. A system of nuclear magnetic resonance (NMR) field probes is used to determine the actual corkscrew trajectories required for a successful and artifact-free wave-CAIPI reconstruction. The utility of GRAPPA-based wave-CAIPI is investigated at a 7 Tesla scanner on the example of two widely-used fast 3D structural MRI methods: T1 weighted gradient echo (MP-RAGE) and T2 weighted spin echo (TSE). The additional spatial information gained with wave-CAIPI sampling allows to significantly increase image quality and SNR of rapidly acquired images. 16-fold accelerated whole brain wave-CAIPI MP-RAGE and wave-CAIPI TSE data with 1 mm isotropic resolution and good image quality are acquired in only 40 seconds and 1:32 minutes, respectively.
The second contribution of this thesis addresses the monitoring and correction of magnetic field fluctuations induced by the patient's physiology. Field perturbations caused by deep breathing or limb motion effect the signal encoding and lead to artifacts in the reconstructed brain images. Nevertheless, many of the image artifacts caused by magnetic field distortions can be corrected if the magnetic field changes are known. Therefore, a field correction approach that accounts for field changes of up to first-order spatial expansion is incorporated into the GRAPPA-based parallel imaging reconstruction. Considering technical limitations of the NMR field probes, such as the minimal time between successive field probe excitations, two field monitoring approaches with different temporal resolutions are investigated and compared for high-resolution T2* weighted 3D-EPI acquisitions at 7 Tesla. Especially for acquisitions that are subject to strong field changes, the temporal SNR strongly benefits from field correction and high-quality images can be regained.},

url = {http://hdl.handle.net/20.500.11811/8442}
}

Die folgenden Nutzungsbestimmungen sind mit dieser Ressource verbunden:

InCopyright