Louis, MO, USA). Creatine was added to deionized water first to reach concentrations of 100 and 125 mM. Once the creatine had fully dissolved, agarose learn more was added to form 3% of the mixed solution and then heated to boiling. After that, the mixed solution was maintained at 50 °C and titrated to pH values of 5.5, 6 and 6.5 before being transferred to different 2 ml vials. A plastic container was used to house all the vials and filled up with agar to minimize field inhomogeneity. The phantoms were left to solidify at room temperature prior to the MRI experiment. All the images were acquired

using a 4.7 T Varian DirectDrive™ spectrometer (Agilent Technologies, Santa Clara, CA, USA). The main magnetic field (B0) was shimmed to minimize field inhomogeneity artifacts and the RF

field was calibrated before experiments. The pulsed parameters used were identical to the simulation: 50 Gaussian pulses, FA = 180°, Tpd = 40 ms, DC = 50% and saturation frequencies from −3.8 to 3.8 ppm (0.19 ppm increments). Crusher gradients with alternating signs were applied after each irradiation pulse to spoil the residual transverse magnetization. A single-slice spin-echo MK 2206 (SE) echo planar imaging (EPI) readout was used at the end of the saturation, with a field of view (FOV) of 80 mm × 80 mm, matrix size of 64 × 64, slice thickness of 1 mm, bandwidth of 250 kHz, echo time (TE) of 20 ms and repetition time (TR) of 4 s. An unsaturated scan with the same image properties was also acquired as a reference. The CEST data were acquired in about 6 min. Besides CEST imaging, relaxation time and magnetic field maps were obtained to account for the inhomogeneity in the scan.

An inversion recovery sequence with eight inversion intervals from 100 to 6000 ms was used to measure the T1 relaxation time of the water pool. Six separate SE images with TEs from 23 to 100 ms were measured to determine the T2 relaxation time of the water pool. The T1 and T2 maps of the water pool were obtained by least square fitting of the image intensity against the TI and TE, respectively. WASSR Fossariinae was applied to find the main magnetic field inhomogeneity. The acquisition parameters were the same as for the CEST imaging, except that the FA was set to 61°. A B0 map was generated by first finding the saturation frequency that recorded the lowest magnetization, then seven saturation frequencies below and above the minimum point were interpolated to intervals of 0.0019 ppm (0.38 Hz). The water center frequency shift was determined using the Maximum Symmetry algorithm [28] based on the interpolated data. The saturation frequency at which the magnetization was minimum was used as the initial value for the search. All the maps and in vitro CEST data were processed using nonlinear least-square curve fitting function, lsqcurvefit in MATLAB (Mathworks, Natick, MA, USA). A three-pool model, which consisted of water (w), amine (labile) and MT, was used to fit the collected data.