TY  - GEN
TI  - Simulation and development of RF resonators for preclinical and clinical 1H and X-nuclei MRI
UR  - https://archiv.ub.uni-heidelberg.de/volltextserver/26315/
AV  - public
A1  - Malzacher, Matthias
CY  - Heidelberg
Y1  - 2019///
ID  - heidok26315
N2  - Magnetic resonance imaging (MRI) is a unique imaging modality since it provides high soft tissue
contrast without ionizing radiation. MRI hardware technology has evolved towards high channel
receive (Rx) systems and highly efficient and homogeneous transmit (Tx) coils both increasing
image quality and reducing measurement time. Radio frequency (RF) systems for MRI are designed
to provide highest signal-to-noise ratio (SNR) while keeping the specific absorption rate (SAR) as
low as possible. Due to the complexity of today?s RF systems for MRI there are still some unsolved
issues, such as preamplifier coupled noise, double resonant clinical coil performance and SAR
simulation accuracy. This thesis consists of three main parts: First, the improvement of the SNR
of Rx systems by evaluating new matching strategies; second, the development of RF setups for
clinical sodium MRI applications; and third, the evaluation of the accuracy of SAR calculations and
electromagnetic (EM) simulations. In the first part a transmit-only receive-only (ToRo) coil setup
was built for 35Cl MRI at 9.4 T. The effect of SNR degradation due to coil coupling was verified
for the Rx array consisting of three Rx coils. A 15 -17% SNR improvement could be achieved
using different matching methods compared to the conventional matching strategy. The combined
SNR of the Rx array was up to 4.5 higher compared to the reference SNR image acquired with the
volumetric Tx coil. In the second part, two different double resonant RF setups for clinical sodium
and proton MRI of the abdomen and the head were realized. A 16 channel abdominal sodium
Rx array was simulated and built to be used with an asymmetric whole-body sodium coil. The
setup was additionally equipped with a local proton transceiver coil. Sodium SNR was improved
about a factor 3 to 6 in phantom measurements compared to the volume coil. The feasibility of
the double resonant setup was proven by an in-vivo sodium and proton scan. The head RF setup
comprised 8 sodium and 8 proton Rx coils. The findings from the 35Cl setup for the avoidance
of SNR degradation due to coupled coils were also applied for this setup. The performance of
the coil was compared to commercial sodium and proton setups using phantom measurements.
Comparable SNR results to the commercial options were achieved in sodium measurements (Rx
array/commercial = 1.14) and comparable SNR (Rx array/commercial = 1.09) as well as parallel
imaging performance (Rx array/commercial: mean g-factor = -7% to +6%; maximum g-factor =
-21% to 33%) in proton measurements. Finally, the feasibility of the coil was proven by in-vivo
sodium and proton measurements. In the third part, the impact of Rx arrays on SAR calculations
was evaluated by modeling a clinical setup at 3T and a research setup at 7T. The clinical setup
comprised a large Birdcage Tx coil and different Rx arrays tailored for the head (24 elements),
the abdomen (36 elements) and the spine (32 elements). The research setup consisted of a head
Birdcage coil and a 32 element Rx array. SAR simulations were performed with and without the
Rx arrays. Mean SAR differences were found to be between -4% and 2% for the 3T setup and up
to 11% for the 7T setup. Maximum SAR differences were found to be between -10% and +6%
for the 3T setup and up to -8% for the 7T setup. As a last step, the accuracy and performance
of the two most common approaches for solving EM fields for MR RF setups were evaluated.
Therefore a whole-body Birdcage coil at 3T, a head Birdcage coil at 7T and an 8 channel Tx array
at 9.4T, 10.5T and 11.7T were modeled. Each setup was simulated using the time domain (TD)
and the frequency domain (FD) method. Finally, the resulting S-parameter, B-fields and SAR were
compared. A difference below 20% was achieved for all these results which was already reported
in earlier studies of a single specialized setup. The FD solver was found to be up to 12 times faster
than the TD solver.
ER  -