To make spectroscopic imaging clinically feasible, rapid and robust acquisitions with high SNR are necessary. We develop and apply rosette spectroscopic imaging at 7T using a 8x2 transceiver array and high degree B0 shimming to acquire rapid (<3min) whole plane brain studies at 0.7cc to 0.3cc resolution. To achieve high spectral bandwidth with moderate gradient demands, two temporal interleaves are used. We demonstrate the performance of this acquisition in controls and tumor patients, with use of regression statistics for determination of abnormality.
A Siemens whole body 7T 8 channel multiple transmit system with body gradient coil, a very high order shim insert (VHOS, Resonance Research Inc.) and 8x2 transceiver array was used. The transceiver array was driven in coil pairs using 8 one-to-two splitters with independent reception from all 16 channels. B1 shimming was performed targeting two distinct RF profiles: the large homogeneous distribution of all intracranial tissue and a ring distribution targeting the superficial skin and skull. B0 shimming was optimized over the slice using BOLERO (1) with high degree shimming.
A planar rosette sequence used a 9mm slice selection and a semi-selective frequency refocusing pulse for water suppression supplemented with a narrow band adiabatic inversion pulse. Lipid suppression was achieved using two adiabatic inversion pulses applied through the spatially distinct B1 ring distribution (2). A moderate TE of 40ms was chosen to reduce macromolecular contributions to the baseline and reduce spectral overlap with amino acids. The spatial encoding was performed using interleaved circular (radial and angular frequencies w1=w2) rosette trajectories with Nshots=52, each shot rotated by 2pi/Nshots. The kmax = Nx/(2*FOV) and spectral BW sampling is achieved by w1 = pi*BW/nTI. To achieve a spectral BW=2500Hz at 7T, multiple temporal interleaves nTI>1 are used; with nTI=2, a single circular trajectory is acquired in 800us (2/spectral bandwidth BW). With temporal interleaves, a relatively mild gradient slew rate of <60mT/m/ms and Gmax of 7mT/m is accommodated; no additional eddy current corrections or trajectory corrections were needed. Reconstructions were performed using a 3D grid (kx, ky and time) with a Kaiser-Bessel kernel with a window W=4 on a 2-fold oversampled grid (3). Phasing and coil combination was performed using a matched anatomical scout. Spatial filtering was applied with a hamming filter in-plane, apodization and convolution difference applied in the time domain prior gridding. LCModel was used for curve fitting.
N=4 control subjects were studied to define the regression statistics for Cr/NAA and Ch/NAA as a function of fraction GM. Tissue segmentation was performed using Freesurfer and fraction GM calculated after convolution of the segmented images with the point spread function of the rosette acquisition and slice excitation profile. LCM spectral analysis was performed evaluating planar frontal and parietal regions and Cr/NA values regressed against fraction gray matter (fGM). Pixel rejection criteria were: total brain content <40%, Cramer Rao values >15%, linewidth >0.12ppm, and maximal excursion due to lipid was <2.0.