MRI of the Hip at 7T using RF shimming with 4-channel excitation Cem Murat Deniz 1,2 , Ryan Brown 1 , Leeor Alon 1,2 , KellyAnne Mcgorty 1 , Hans-Peter Fautz 3 , Graham Wiggins 1 , Daniel K. Sodickson 1 , Yudong Zhu 1 , Riccardo Lattanzi 1 1 Center for Biomedical Imaging, Department of Radiology, NYU School of Medicine, New York, NY, United States, 2 Sackler Institute of Graduate Biomedical Sciences, NYU School of Medicine, New York, NY, United States, 3 Siemens Medical Solutions, Erlangen, Germany Introduction: The feasibility of high-resolution in vivo imaging of the musculoskeletal system at 7T has been demonstrated for superficial joints such as the wrist (1) and knee (2). However, ultra-high field MRI is more difficult in deeper joints in the body, such as the hip, due to B 1+ inhomogeneity and specific absorption rate (SAR) limitations. The objective of this work is to show the feasibility of MRI of the hip at 7T using multi-channel transmission. For comparison, the signal-to-noise ratio (SNR) in the hip articular cartilage was measured at 7T and 3T. Methods: 7T experiments were performed on a Siemens whole body Magnetom scanner (Erlangen, Germany) equipped with an eight-channel parallel transmit system. The institutional review board approved the study, and informed consent was obtained from the volunteer. A 10-channel array consisting of five loop/stripline modules (3) was wrapped around the left pelvis. Four loop coils (15cm x 8cm) were used for RF transmission and reception, while the additional elements were used for RF reception only. In vivo RF power limits were estimated from the RF power required to heat a body-size agar phantom with uniform electrical properties of average human muscle (electric conductivity = 0.77 S/m, dielectric constant = 58) at a rate of 1°C per 10 mins. Individual transmit channel B 1+ calibration was performed following the method described in Ref. (4). Using calibrated flip angle (FA) maps, subject dependent local transmit B 1+ shimming was used to obtain maximum transmit efficiency in the left hip articular cartilage. Following the notation of Ref. (5), we determined maximally-efficient B 1+ shim weights by finding the eigenvector associated with the largest eigenvalue of the following transmit efficiency metric: / H H   w Γw w Ιw , where w is the vector containing B 1+ shim weights, I is the identity matrix and Γ is the matrix containing average B 1+ strength square of the unit shim weights. Calculating maximum of η can be formulized as a generalized eigenvalue problem. For convenience, we used the identity matrix I in the denominator of this expression, in place of the power correlation matrix defined in Ref. (5). This substitutes the sum of squares of shim amplitudes for the total RF power dissipation, which may be obtained if desired using additional experimental calibration (6,7). In order to compare SNR, 3T experiments were performed on a Siemens Verio MR system using the body coil for RF transmission and a 32- channel cardiac coil array (Invivo Corp.) wrapped around the subject’s pelvis for RF reception. The same subject was imaged at both 3T and 7T. Axial GRE images in the hip joint were acquired at 3T and 7T with the following parameters: acquisition matrix = 512x512, spatial resolution = 0.6x0.6x2 mm 3 , TR/TE = 400/4.73 ms, receiver bandwidth = 300 Hz/pixel, transmit voltage = 357/80 V (3T/7T). SNR maps were generated following the method of Kellman et al. (8). For fair comparison, normalized SNR maps were obtained by dividing the SNR maps by the sine of the FA at each voxel. Results: For 7T experiments without targeted B 1+ shimming, B 1+ inhomogeneity resulted in large signal intensity variations throughout the field of view and especially in the hip region, as indicated by the red arrow in Fig 1a. A region of interest surrounding the left hip articular cartilage was selected as the target region for maximum efficiency B 1+ shimming (Fig. 1b). Figure 1c shows improved homogeneity in the hip region as a result of B 1+ shimming. The shim also resulted in a 56% increase in transmit efficiency in the targeted region of interest. Axial GRE images, and SNR and FA maps overlaid on GRE images are shown in Figure 2. FA's of only 23° ± 9.5° (mean ± SD) were generated in the hip cartilage at 7T (Fig. 2d) because of the conservative parallel transmit power limits imposed in this study, while FA's of 83.6° ± 6.6° were generated at 3T (Fig. 2c). Despite the reduced FA at 7T, the raw SNR was greater: 23.3 ± 10.5 at 7T versus 17.9 ± 6.1 at 3T (Fig. 2e,f). Even in the deepest medial region of the hip cartilage, where SNR was lowest at 7T, a SNR gain of ~35% over 3T was observed. SNR normalized by the FA was 18.2 ± 6.3 at 3T and 53.1 ± 25.6 at 7T. Discussion: The feasibility of MRI of the hip at 7T was demonstrated using multiple-coil transmission with B 1+ shimming. The normalized SNR was 2 to 3 times greater at 7T compared to 3T, showing the benefit of higher fields for hip imaging. One limitation of this comparison is the difference in receive coil sensitivities at 7T and 3T due to variation in coil size and structure. However, the hip lies at a similar depth (5 to 12 cm from the body surface) as the heart, so that the cardiac array used at 3T may serve as a reasonable SNR benchmark. In this study, conservative SAR limits were imposed at 7T. We expect larger gains in raw SNR by incorporating SAR calibration methods (6,7) that may permit relaxed power limits and allow higher FA. References: [1] Chang G. et al. (2010): JMRI 31: 740-6. [2] Regatte RR. et al. (2007): JMRI 25: 262-9. [3] Brown R. et al. (2010): ISMRM: 3807. [4] Fautz, H-P et al. (2008) ISMRM 1247. [5] Zhu Y. et al. (2010): ISMRM 1518. [6] Alon L, et al. (2010): ISMRM 780. [7] Zhu Y. (2009) ISMRM: 2585. [8] Kellman P. et al. (2005): MRM 54: 1439-47. Figure 2: a,b: Axial GRE images at 3T (left) and 7T (right). c,d: Measured flip angle maps in the hip cartilage. e,f: SNR maps before flip angle normalization Figure 1 a: 7T axial GRE image without B 1+ shimming. b: Selected region of interest for shim weight calculation. c: Axial GRE image with B 1+ shimming No Shim Shim ROI Shimmed a b c 7T 3T a b 100 0 55 0 c d c d e f