Ultra-High-Resolution Imaging of the Human Brain at 9.4 T Using k-Space Weighted Acquisition

Introduction: Acquisition-weighted (AW) imaging by acquiring a varying number of averages depending on the position in k-space is widely used in CSI for suppression of contamination by the sidebands of the spatial response function (SRF). In addition, it has been shown that the apparent SNR can be increased due to the reduction of negative signal contributions from adjacent voxels [1]. In cases where averaging is required anyway for SNR reasons, AW can be realized without sacrificing spatial resolution or scan time. So far, these advantages have only rarely been used in imaging [2]. For ultra-high resolution imaging, SNR and Gibbs-ringing are crucial, and averaging is often used to obtain sufficient SNR. Here, this technique is used at a field strength of 9.4 T with a highly sensitive receive array to obtain images from the human brain with a voxel size down to 14 nl. Methods: Data was acquired at a 9.4 T scanner on human subjects, using a 16 channel transmit/31 channel receive coil combination [3]. 3D GRE images were acquired with a conventional sequence with 3 averages and an AW technique with the same scan time, with weighting in both PE-directions based on a Hanning function, but modified for the relatively low number of scans [4]. The resolution was kept equal by covering a larger region of k-space in the weighted image, which was verified by an analysis of the width of the SRF. First images with a resolution of 0.2×0.2×0.5 mm were acquired within 13:20 min with both techniques (TE 18.4 ms, TR 26 ms, 16 slices, BW 120 Hz/px, FA 10°). Ultra high resolution phase images (0.13×0.13×0.8 mm) were processed from an AW data set (TE 17.3 ms, TR 27 ms, 22:09 min).