Radiofrequency bandwidth effects in direct saturation dynamic glucose enhanced MRI at 3T

Direct Saturation Dynamic Glucose Enhanced (DS-DGE) MRI is a novel technique that utilizes low B₁ strength and short saturation pulse duration to measure the effect of enhanced T₂ relaxation caused by the infusion of D-glucose, resulting in linewidth broadening of the direct water saturation (DS) signal in the water saturation spectrum (Z-spectrum). In this simulation-supported protocol optimization study, we investigated this approach for brain tumor applications, with particular emphasis on the design of the radiofrequency (RF) saturation pulses. Artifacts related to the bandwidth of the RF saturation pulses, affecting the linewidth derived from Lorentzian fitting of the DS resonance, were investigated.

In vivo DS-DGE MRI Z-spectra were analyzed using Lorentzian line-shape fitting. To support experimental findings, simulated Z-spectra were generated using Bloch-McConnell simulations implemented in Pulseq-CEST, covering a wide range of T₁ and T₂ relaxation times. Diverse RF saturation schemes were employed, varying in saturation duration, number of pulses, and pulse shape. The accuracy of Lorentzian fitting was assessed across these RF pulse configurations.

RF pulses for a fixed total duration of 500 ms, with maximized individual pulse length, resulted in less sideband artifacts compared to using more pulses of shorter individual length. Choosing fewer than 5 pulses yielded Lorentzian linewidth estimates with a root-mean-square-error (RMSE) below 1 %.

The use of longer RF pulses, within the 500 ms total saturation duration, improved estimation accuracy when employing a Lorentzian fit. This approach can be readily implemented without major modifications to the acquisition or post-processing protocol. These findings therefore provide a practical basis for future in vivo DS-DGE studies, as well as for other applications employing linewidth-sensitive Z-spectra approaches.