Comparison of k‐space sampling schemes for multidimensional MR spectroscopic imaging

For clinical 31P MR spectroscopic imaging (MRSI) studies, where signal averaging is necessary, some improvement of sensitivity and spatial response function may be achieved by acquiring data over a spherical k-space volume and varying the number of averages acquired in proportion to the desired spatial filter. Eight different k-space sampling schemes are compared through simulations that provide graphs of the spatial response functions (SRF), and tabulations of voxel volumes, relative signal-to-noise ratios (SNR), and relative data collection efficiencies (SNR per unit volume over the same time). All schemes were based on practical experiments, each of which could be implemented in the same length of time. The results show that in comparison with cubic k-space sampling with the same number of signal averages at each point, spherical and acquisition-weighted k-space sampling can be used to achieve reduced Gibbs ringing along the principal axes directions, and thus reduced contamination from adjacent tissue in these directions, without degradation of voxel volume or SNR.