Abstract
Functional (31)P MRS might give valuable information on the energy metabolism of the brain during stimulation. However, the tiny expected spectral changes are difficult to detect due to low SNR and large voxel sizes. Higher field strengths and sensitive multielement arrays could help to investigate the energy metabolism during brain activation. Here, we acquired functional (31)P MRS data during visual stimulation at 9.4 T using a 27-element receive array and an optimized 3D-CSI protocol. First, T(1) values at that field strength were measured, confirming a decrease in T(1) times for increasing fields. Then, a 4.5 min visual stimulus with varying color scheme and frequency was presented to healthy subjects during CSI acquisition for 45 min. Stimulus and rest signals were formed by averaging over five epochs, resulting in spectra with high spectral quality and SNR. Stimulus and rest spectra were almost identical, except for a very small but significant difference in the chemical shift of inorganic phosphate, which might indicate a slight increase in pH during stimulation. It was then investigated whether postprocessing algorithms might be able to detect more subtle changes. Spectral denoising using a principal component approach resulted in improved SNR with high spectral integrity but suppressed the chemical shift difference. Spatial deconvolution improved the spatial resolution but did not yield changes in the outcome. Those results indicate that only very subtle changes in the (31)P spectrum due to visual stimulation can be expected, which are difficult to detect even when taking advantage of the high SNR and spectral dispersion at 9.4 T.