Glutamate-Weighted CEST Contrast After Removal of Magnetization Transfer Effect in Human Brain and Rat Brain with Tumor.

Abstract

To mitigate the effect of magnetization transfer (MT) from glutamate-weighted chemical exchange saturation transfer (GluCEST) contrast in healthy human brain and its demonstration in a rat brain tumor model. GluCEST data was acquired from six healthy human volunteers at 7T and on a rat brain with tumor at 9.4T. Single voxel proton magnetic resonance spectroscopy (1HMRS) data was acquired from three human volunteers. The magnetic resonance imaging protocol included CEST data acquisition at multiple frequencies for generating Z-spectra, B0 and B1 map. Partial Z-spectra at offset frequencies from ± 100 to ± 14ppm were fitted to model semi-solid MT component by Lorentzian, Gaussian, super-Lorentzian, and 6th degree polynomial function lineshapes. Average residual errors per pixel was calculated. The MT effect of the Z-spectra was removed by subtracting fitted MT component from Z-spectra. GluCEST was computed as GluCESTNeg (normalized with signal at - 3ppm) and GluCESTM0 (normalized with unsaturated signal). The difference between GluCEST maps before and after MT removal was compared using T test. Better accuracy of fitting off-resonance Z-spectra was achieved with super-Lorentzian (σ = 0.0009) and Lorentzian (σ = 0.0017) compared to other lineshapes. There was significant (p < 0.01) increase in GluCESTM0 and decrease in GluCESTNeg contrast after MT removal. GluCESTNeg and GluCESTM0 maps after MT removal using Lorentzian lineshape showed gray matter (GM) to white matter (WM) contrast ratio of 1.47 and 1.52 respectively. These ratios are close to glutamate concentration ratio in GM/WM as observed from 1HMRS data. Thus, the quantity of the MT removed from Z-spectra is appropriate using Lorentzian lineshape due to preservation of GluCEST contrast-ratio in GM/WM. The amount of MT removed from Z-spectra is overestimated using super-Lorentzian and underestimated using Gaussian and polynomial lineshapes. Tumor tissue showed unexpected increase in GluCEST contrast compared to contra-lesional tissue, which represents normal appearing tissue in the brain on contra-lateral side of tumor region, due to decrease in MT component. Removal of MT effect from Z-spectra using Lorentzian lineshape increased the specificity of GluCEST contrast to glutamate in healthy human brain and was demonstrated in rat brain tumor model.