Multi-Band MRSI at 7T using 3D B1 Shimming based Outer Volume Suppression
Hoby Patrick Hetherington1, Tiejun Zhao2, Victor Yushmanov1, and Jullie Pan3

1Radiology, University of Pittsburgh, Pittsburgh, PA, United States, 2Siemens Medical Systems, New York, NY, United States, 3Neurology, University of Pittsburgh, Pittsburgh, PA, United States


To provide near whole brain coverage for both anatomical imaging and MRSI we used an 8x2 transceiver array with 8 independent RF channels and eight 1 to 2 splitters. This configuration provided a homogeneous RF distribution (<12% SD, 750Hz peak B1) while enabling 3D RF shimming based outer volume suppression to minimize extra-cerebral lipid signals. MRSI data was acquired at 7T from control subjects and patients with mTBI with a multi-band MRSI sequence (four simultaneous slices) using two RF distributions. Increases in choline/NAA were seen in both the anterior frontal lobe and the hippocampi.


Achieving adequate B1 amplitude, homogeneity and coverage are critical components for MRSI at 7T. Previously we demonstrated that RF shimming with an 8 element transceiver array enables peak B1 amplitudes of a 1.0 kHz at moderate peak total powers (2-4kW) and RF shimming based outer volume suppression for MRSI[1]. However, to maintain high efficiency, an 8 element transceiver array is inherently limited and does not provide simultaneous homogeneous whole brain coverage. Further, due to limited homogeneity in the axis along the coil, RF shimming based outer volume suppression typically requires multiple RF shim values. Combined these factors make multi-slice or 3D acquisitions challenging. To extend the coverage for both anatomical imaging and MRSI we used an 8x2 transceiver array [2] (Fig. 1A) that provides homogeneous whole brain coverage for multiband excitation and enables 3D RF shimming based outer volume suppression to minimize contamination from extra-cerebral lipids.


All data were acquired on an 8 channel parallel transmit 7T Siemens human MRI system using an 8x2 transceiver array and a higher order shim insert (Resonance Research Inc.). Using 8 independent RF channels the 16 coil transceiver array was driven using eight 1 to 2 splitters. In this configuration (Fig 1B), each independent RF channel drives two coils (same “column” across 2 rows). A homogeneous (Fig 2A) and ring distribution (Fig 2B) are achieved using RF shimming without gradients. A multi-band MRSI acquisition (Fig 3;TR/TE=1500/30ms) was used to excite 4 contiguous slices (8mm thick, 2mm gap) that are un-aliased using the sensitivity profile of the individual coils. Four cascaded RF excitation pulses are used to provide high peak B1 and minimize chemical shift registration artifacts. The phase of each excitation pulse is toggled by 180 degrees in phase depending upon the values of kx and ky to distribute the aliased signals from each slice in both phase encoding directions (Fig. 3). This minimizes slice overlap and contamination. The excitation and refocusing pulses were driven with the homogeneous distribution. A single broad-band semi-selective refocusing pulse is used for water suppression. Outer volume suppression is provided by a double inversion recovery sequence using the ring distribution.


Data were acquired from 5 control subjects and 5 subjects with mild traumatic brain injury (mTBI). Using the homogenous distribution a B1 amplitude of 750Hz was achieved (11.8% SD – Fig 2A) over the entire cerebrum using less than 160V per channel (before splitting). The ring distribution was achieved using less than 80V per channel. Displayed in Fig. 4 are representative MRSI data from a healthy control and an MRI negative subject with mTBI. The spectrum from frontal gray matter demonstrates a large increment in the choline/NAA ratio, consistent with axonal and neuronal injury, despite being MRI negative. For mTBI, where the site of injury is variable and cannot be routinely identified by clinical MRI, providing extended brain coverage for MRSI is critical. To further evaluate the extent to which a single set of RF shim values (I.e. those used for the MB-MRSI) could also be used for more inferior locations and oblique slices we also acquired single slice MRSI data along the hippocampi. Despite the use of an oblique slice orientation excellent spectral quality and lipid suppression was obtained from the hippocampi (Fig. 5). Elevations in choline/NAA were also detected from anterior hippocampal regions in the subjects with mTBI consistent with previous reports in veterans with mTBI [3].


The 8x2 transceiver array provides for near whole brain homogeneous coverage and RF shimming based outer volume suppression. The spatial extent of the homogeneous and ring distributions enables multi-band excitations to be used for MRSI. Using phase toggling to distribute the aliased data in two dimensions enables a multiband factor of 4 to be used while preserving spectral quality. We have used these methods to identify areas of axonal and neuronal injury (increased choline/NAA) in MRI negative subjects with mTBI.


NIH R01-NS081772, R01-EB011639, R01-NS090417


1. Hetherington, H.P., et al., RF shimming for spectroscopic localization in the human brain at 7 T. Magn Reson Med, 2010. 63(1): p. 9-19.

2. Avdievich, N.I., Transceiver-Phased Arrays for Human Brain Studies at 7 T. Appl Magn Reson, 2011. 41(2-4): p. 483-506.

3. de Lanerolle, N.C., et al., Concussive brain injury from explosive blast. Ann Clin Transl Neurol, 2014. 1(9): p. 692-702.


A: 8x2 transceiver array; B: RF drive scheme and splitter

A: Scout images; B: B1 maps homogeneous distribution; C: B1 maps ring distribution; D: MP2RAGE images

Mullti-band MRSI sequence. Four simultaneous slices are excited with the homogeneous RF distribution. Outer volume suppression is provided by a double inversion recovery sequence with the ring RF distribution. Overlap of the 4 slices is depicted (green - no overlap, yellow 2 slices overlap, orange -3 slices overlap)

3rd slice of 4 slices from a MB-MRSI data set from a patient with mTBI and a control. Displayed are the scout image, NAA image and a spectrum, (green box on each image). Some residual lipid remains from the scalp, however this does not substantially contaminate voxels inside the head.

Single slice data from the hippocampus using the same coil and B1 amplitude and phase parameters as the MB-MRSI. MRSI data from mTBI and control subject. A large increase in Ch/NAA is seen in the mTBI subject in the anterior hippocampus (green box on image).

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