Introduction

Magnetic resonance elastography (MRE)¹ has been found to provide important information about various brain diseases^1-5. Implementing MRE on an ultra-high field scanner may provide higher signal-to-noise ratio (SNR) and allow for the higher spatial resolution needed for brain MRE. However, ultra-high field also presents challenges such as longer T1, shorter T2 and increased B1 inhomogeneity^{6, 7}. Braun et al. tested high resolution brain MRE at 7T with promising results⁶. The goal of this study was to compare MRE image quality, including stiffness measurements and SNR, with images acquired at 7T and 3T using similar imaging sequences and protocols.

Methods

A PVC head phantom was used in this study (Figure 1). Images were acquired using prototype pulse sequences on a clinical 7T scanner (MAGNETOM Terra, Siemens Healthcare, Erlangen, Germany) and a clinical 3T scanner (MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany). At 7T dielectric pads were added to both sides of the phantom as shown in Figure 1⁸. At each field strength, a soft passive driver as shown in Figure 1 was placed under the phantom to induce motion in the phantom. On each scanner, the motion amplitude was matched, the vibration frequency was 60Hz, and eight phase offsets were acquired. For the purpose of comparison, a prototype spin-echo EPI-MRE sequence was developed for the 7T scanner mirroring as close as possible the sequence used at 3T^9-12. The parameters were as comparable as possible to each other: FOV= 240x240mm², 20 slices were acquired with a slice thickness of 2mm. The matrix size was 128x128 (voxel size 1.88x1.88x2mm³). TR/TE=4000ms/65ms at 7T and 4000ms/70ms at 3T. The bandwidth at 7T was 2170Hz/pixel and at 3T 1860Hz/pixel. The GRAPPA factor was 3. For both protocols, two motion encoding gradient cycles were used with the gradient amplitude of 10mT/m. Three motion encoding directions were acquired.
For the 3T cases composite magnitude and phase difference images were reconstructed by combining the data acquired with 2 opposite motion encoding polarities. For 7T, the real and imaginary images were reconstructed, and magnitude and phase images were calculated offline. For SNR estimation, regions of interest (ROI) for signal and noise were drawn on the magnitude images of the center slice of the 4th phase offset. Octahedral shear strain based SNR (OSS-SNR)¹³ was estimated based on a 3D ROI of the whole phantom.

Results

Figure 2 shows the real and imaginary images acquired at 7T and the magnitude and phase difference images from off-line reconstruction. Figure 3 shows the phase difference images, curl maps and stiffness measured in the 3T and 7T images, respectively. Note that the phase difference images and curl maps acquired at 7T are less noisy than those at 3T. Stiffness in the center slice was 2.72 kPa when estimated from images acquired at 7T and 2.51 kPa when estimated from images acquired at 3T. The SNR estimated from the 7T magnitude image was 58.9, while from the 3T composite magnitude image, the estimated SNR was 35.7. The OSS-SNR from wave images was 3.39 for 7T and 1.45 for 3T.

Discussion

In this study 3D MRE performed using similar imaging sequences and protocols at 3T and 7T was compared with regards to estimated stiffness values, SNR and OSS-SNR. Our results showed that the estimated stiffness values were comparable between 3T and 7T. The image SNR was much higher at 7T; and the OSS-SNR more than doubled at 7T scanner compared with that at 3T. These results show the potential for brain MRE with high resolution or low motion amplitude at 7T, which could either help detect smaller lesions in the brain, or improve patient experience when using MRE. Future work includes a volunteer study and testing higher spatial resolution and different vibration frequencies.

Acknowledgements

This work was supported by grant from National Institutes of Health R01 EB001981.