Prospect of SNR and SAR Improvement on a Whole-body Human 10.5T Scanner using High Dielectric Material
Sebastian Rupprecht1, Hannes M Wiesner2, Pierre-Francois van De-Mortelle2, Byeong-Yeul Lee2, Wei Luo3, Xiao-Hong Zhu2, Isaiah Duck1, Gregor Adriany2, Christopher Sica1, Kamil Ugurbil2, Michael Lanagan3, Wei Chen2, and Qing Yang1

1Department of Radiology, The Pennsylvania State University College of Medicine, Hershey, PA, United States, 2Radiology Department, Center for Magnetic Resonance Research, Minneapolis, MN, United States, 3Department of Engineering Sciences and Mechanics, The Pennsylvania State University, State College, PA, United States


We compared and characterized the RF field wave behavior for human brain imaging at 10.5T and 7T. Additionally we explored the feasibility of using monolithic high dielectric constant materials to potentially further enhance SNR and circumvent SAR limitations and show that there can be great benefits through phantom experiments and computer modeling.


The installation of the world first 10.5T human MRI system at the University of Minnesota, the Center for Magnetic Resonance Research (CMRR) once again, pushes the boundary of human MRI technology into an uncharted territory. With the increased field strength, prominent wave behavior exhibited at 7T is expected to be even stronger, which could adversely impact anticipated SNR benefits and (local and global) SAR limitations at the higher field strength. Thus, in order to seize the full potential of this unique instrument for human brain research, it is necessary to first carefully examine the RF behavior in the human head at this new field strength. The wave behavior seen at high fields stems from the dielectric properties of the human body, which lead to the wave propagation patterns at 300 and 447 Mhz. In fact, the dielectric effect originally inspired the use of high dielectric constant (HDC) materials to manipulate the wave behavior in an attempt to increase SNR and decrease SAR with monolithic high dielectric constant materials as has been previously shown at lower frequencies1-3. This study characterized RF wave behavior at 10.5T in human brain model and phantom in comparison to that at 7T and explored the potential for even greater improvement of SNR using high dielectric materials at 10.5T.


Figure 1a shows the baseline computer modeling setup, consisting of a human head model (Virtual Family Ella) with an 8 cm diameter T/R surface coil driven by unit current sources. The simulation was performed at 300 MHz for 7T and 447 MHz for 10.5T using xFDTD (REMCOM, Inc., PA). At 10.5T a parameter sweep for a dielectric disk (placed between the head and the coil as shown in Fig. 1b) was run to determine an approximate starting permittivity for this coil within our manufacturing constraints. A disk (diameter 80 mm, thickness 10 mm) with a permittivity of 90 was selected as it both strongly and homogeneously enhanced B1+ up to our desired depth of about 5 cm. This disk was then manufactured, characterized and finally used for the experimental part of the study. Additionally, we constructed an RF coil (ID 80 mm), which was placed in a custom coil holder (drawer like for coil height adjustment, see Fig. 1c). For the baseline the coil was as close as possible to the phantom (8.2g Na C2H3O2 + 9.6g C3H5O3Li per L). This phantom was placed on top of the setup as shown in Fig. 1d. Once the dielectric was placed between coil and phantom the coil was moved about 1 cm away from the phantom to make room for the disk. Capacitor values for tuning and matching were adjusted for each case. Experimentally, relative B1+ (AFI technique4), receive sensitivity and noise scans were obtained on the Siemens Magnetom 10.5T MR-system. Flip angle and RF power were carefully adjusted to compare baselines and HDC cases.

Results and Discussion

As shown in Fig. 2, RF transmission field (B1+) distribution at 10.5T is distinctively different from 7.0T. As a result of shorter wavelength at 10.5T (l10.5T ~ 9.5 cm vs. l7.0T ~ 14 cm in an average brain), the constructive (white arrows) and destructive (black arrows) interferences become more prominent, and the penetration depth appears shorter than that of 7.0T. These results suggest that Transmit SENSE and RF shimming methodologies that are currently being developed at 7.0T can be more effective at 10.5T. Figure 3 shows the comparison between the simulated and experimental results. The line plots of B1+ with and without the HDC pad with respect to the distance from the center of the RF coil at 10.5T. With this configuration, the transmit B1+ field increased from between 66% to 30% and within 1 to 5 cm into the sample, which can be translate into requirement of using less RF transmission power for a given flip angle. These results, although obtained under suboptimal conditions, also suggest that the SNR achievable at 10.5T can be further enhanced significantly by incorporating HDC material with a RF coil. Experimental measurement validated our calculation results, showing approximately 60 to 30% enhancements from 1 cm to 5 cm into the sample.


Our experimental and simulated results demonstrate that stronger wave behaviors associated with the dielectric effect at 10.5T could make RF shimming more effective than at 7.0T. Incorporation of high dielectric material into the RF coil could synergistically improve SNR and RF shimming by reducing the overall necessary RF power to achieve the same flip angle, thus reducing the overall and local SAR5 constrains during RF shimming.


NIH grants of R24 MH106049, RO1 NS070839, S10 RR029672, P41 EB015894 and P30 NS076408.


1 Rupprecht et al. ISMRM 2013

2 Aussenhofer et al., JMR 2014

3 Rupprecht et al. ISMRM 2014

4 Yarnykh, MRM 2007 (57)

5 Sica et al. ISMRM 2014


Fig. 2. B1+ field distribution at 7.0T and 10.5T with and without a HDC disk.

Fig. 3. Numerically calculated (left, Virtual Family Ella) and experimentally measured (right, phantom) B1+ maps and line plots at 10.5T with and without the dielectric disk. The RF surface coil was moved 1 cm away from sample to accommodate the HDC disk.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)