A linear model‐based, B1 inhomogeneity–corrected, glutamate‐weighted chemical exchange saturation transfer MRI at 5 T

Glutamate-weighted chemical exchange saturation transfer (gluCEST) MRI shows promise in the diagnosis and evaluation of neurological disorders. GluCEST must be conducted at high magnetic field strengths, in which pronounced B1 inhomogeneity compromises gluCEST measurements. Conventional B1 inhomogeneity correction methods rely on interpolation algorithms, B1 choices, acquisition numbers, or calibration curves, making correction unrobust. This study proposes a linear model-based method for B1 inhomogeneity correction in gluCEST MRI. Four healthy volunteers and 4 brain-tumor patients underwent B1 field mapping and chemical exchange saturation transfer (CEST) imaging at three nominal B1 levels of 3.5, 4.0, and 4.5 μT at 5 T. The linear relationship between 1/Z(±3 ppm) and B1 was investigated using numerical simulations and human-brain data. The 1/Z(±3 ppm) values at nominal B1 were calculated based on the linear function constructed from either two or three CEST acquisitions. GluCEST signals, quantified with 1-Z(+3 ppm)/Z(-3 ppm), were compared before and after B1 inhomogeneity correction, including those corrected with the conventional interpolation method. A linear dependency of 1/Z(±3 ppm) on B1 within a B1 range of 2-5 μT was identified. Evident B1 inhomogeneity artifacts were observed in the uncorrected gluCEST images. GluCEST signals corrected using the conventional interpolation method varied with CEST acquisition numbers, particularly in regions with severe B1 inhomogeneity. In contrast, the linear model yielded consistent gluCEST contrasts with less dependence on CEST acquisition numbers, superior to the interpolation method. The proposed linear model allows robust B1 inhomogeneity correction from at least two CEST acquisitions, providing an effective way for improved gluCEST MRI.