Correction of time-dependent phase fluctuations in diffusion-weighted MRS at high b-values with external phantom reference

Diffusion weighted MRS (dMRS) provides information on intracellular cell morphology by measuring the mobility of cell-specific metabolites. The strong diffusion encoding gradients needed for dMRS generate eddy currents and other field fluctuations like acoustic vibrations that can severely distort spectral line shapes. Therefore, experimental optimization and corrections with water reference scans are needed. However, in experiments with many and strong diffusion encoding conditions, such as powder-averaged acquisitions , water reference scans can be time-consuming and prone to noise. Other correction approaches require special mapping sequences or advanced field camera equipment that is less effective for the typical 10-100 Hz field oscillations created by diffusion encoding gradients. In this study, we propose the use of a simple phantom measurement that provides high SNR phase reference data as an alternative to a conventional water reference to correct for the spatial characteristics of field fluctuations in a dMRS acquisition on a human 7T system. In simulations, we demonstrate the impact of noise in reference data for phase and frequency correction of low SNR dMRS data. Finally, we compare the performance of the phantom reference with a conventionally acquired water reference scan in a protocol with an exceptionally high b-value of 37 ms/µm².