Evaluation of B0-inhomogeneity correction for triple-quantum-filtered sodium MRI of the human brain at 4.7 T.

Off-resonance can result in signal loss on triple-quantum-filtered (TQF) sodium images. Three correction methods have been proposed to mitigate this problem, but their effectiveness and necessity has not yet been evaluated for human brain. This evaluation is warranted given the doubling or quadrupling of scan length without the expected signal-to-noise ratio (SNR) benefit. First, simulations and agar gel experiments showed that the off-resonance effects on signal loss were asymmetric about on-resonance. Second, the two scan length doubling correction methods were tested for two sets of TQF acquisition parameters in 10 healthy volunteers at 4.7 Tesla. Using only manual shimming on the sodium signal and a 3-pulse TQF sequence with an optimal preparation time value of 6 ms, the majority of brain tissue voxels (87-94% depending on sequence parameters) experienced B0 inhomogeneity amounting to less than 10% signal losses. Relative signal intensities of 0.96 ± 0.04 and 0.98 ± 0.02 were measured in these voxels relative to on-resonant voxels for SNR-optimized and standard TQF parameters. The remaining brain voxels in regions with known susceptibility problems suffered more substantial signal losses, which were partially recovered with the correction methods. At field strengths below 4.7T, at similar ranges of offset frequencies at higher fields and in typical volunteers, B0 correction appears unnecessary for TQF analysis in most of the brain. In many cases where regions with known susceptibility issues are not of concern, a doubling of scan time may be better spent to either improve SNR or spatial resolution in the TQF sodium images.