Spatial
averaging of multiple high-resolution CSI (hrCSI) voxels is commonly employed to
gain SNR and improve quantification of metabolites. Using in vivo 17-oxygen 3D CSI, we compared SNR between spatial averaging
of multiple hrCSI voxels and a single voxel acquired with low-resolution CSI
(lrCSI) with matched sample volume and position. SNR from voxel averaging was much
lower than that of lrCSI caused mainly by the increased noise level by spectral
summation. This
study clearly demonstrates that the acquisition of high-resolution data with
spatial averaging faces a large trade-off of SNR. Therefore, it should be taken
consideration carefully for the choice of an appropriate voxel size of
high-resolution CSI for in vivo study
of neurological or metabolic diseases.
