To put forward a chemical exchange saturation transfer (CEST) imaging scheme based on spatiotemporally encoded (SPEN) MRI to shorten the acquisition time, enhance the immunity to inhomogeneous fields and improve the quality of CEST images.

The CEST-SPEN pulse sequence is shown in Fig. 1. The acquired signal $$$S({t_a})$$$ can be expressed as¹:$$S({t_a}) \propto \int_{ - {L_y}/2}^{{L_y}/2} {\rho (y)\exp [i(\frac{{\gamma {G_{exc}}{T_{exc}}}}{{2{L_y}}}{y^2} - \frac{{\gamma {G_{exc}}{T_{exc}}}}{2}y + \frac{{\gamma {G_{exc}}{T_{exc}}{L_y}}}{8} + \frac{\pi }{2} + \gamma \int_0^{{t_a}} {{G_{acq}}dt \cdot y} )]} dy$$where $$$\rho (y)$$$ is the spatial profile of the spin density, $$$\gamma $$$ is the gyromagnetic ratio, $$${G_{exc}}$$$ and $$${T_{exc}}$$$ are the amplitude and duration of encoding gradient, $$${G_{acq}}$$$ is the decoding gradient and $$${t_a}$$$ is the duration of it. The spatiotemporal encoding is applied along the y-axis with a field of view (FOV) $$${L_{y}}$$$. A discrete form of the data acquisition model is $$$S = \Phi \rho $$$, where $$$\Phi $$$ denotes the quadratic phase modulation. According to the compressed sensing (CS) theory, we can reconstruct ρ by solving the following minimization problem²:$${\kern 1pt} {\kern 1pt} \arg \mathop {\min }\limits_\rho \left\| {S - \Phi \rho } \right\|_2^2{\text{ + }}{\lambda _1}{\left\| {\Psi \rho } \right\|_1}{\text{ + }}{\lambda _2}{\left\| {E \cdot TV(\rho )} \right\|_1}$$where the contourlet transform $$$\Psi $$$ and the total variation (TV) penalty is employed to sparse the images. $$${\lambda _1}$$$ and $$${\lambda _2}$$$ are weighting parameters governing the tradeoff between the reconstruction error and the sparsity. E is the edges of the original image which is used to protect the edges of CEST image and improve the result. The Lorentizan fitting is used to calculate the CEST value. The Lorentizan equation can be described as³:$$L = \frac{{A{\tau ^2}}}{{{\tau ^2} + 4{{(\omega - \delta )}^2}}} + b$$where A is the amplitude of the Lorentzian curve, $$$\tau $$$ is the linewidth of water peak, $$$\delta $$$ is the frequency shift of water peak due to magnetic field inhomogeneity, and b is a global baseline offset on the Z-spectrum caused by the magnetization transfer (MT) effect. The nuclear Overhauser enhancement (NOE) effect is a common effect in in vivo CEST experiments under high field, so NOE images are obtained as a contrast mechanism in this study. The conventional fast spin-echo (FSE) and echo planar imaging (EPI) methods are applied as references to confirm the effectiveness of our scheme.

Experiments were performed on a Varian 7.0 T MRI system using a quadrature-coil probe. The sample was tumor rats. The rats were placed on the bed and anaesthetized safely by using anesthesia machine in which the flowing gas was a mixture of pure oxygen and chlorine halothane with a specific proportion, 5% for quick anesthesia and 2% for maintaining. The parameters for acquisition were set as follows: FOV = 50 × 50 mm², matrix = 64 $$$ \times $$$ 64 (SPEN and EPI) and 128 $$$ \times $$$ 128 (FSE), thickness = 2.0 mm, frequency sweep bandwidth of chirp pulse = 64 kHz, duration of chirp pulse = 3 ms, RF saturation power = 1.47 μT and saturation time = 3 s. In the Z-spectrum experiments, the saturation pulse frequency was swept from -6 ppm to +6 ppm with an increment of 0.5 ppm. For CEST-SPEN MRI and CEST-EPI experiments, the entire acquisition time of a Z-spectrum was 225 s, and the acquisition time of CEST-FSE was 60 min. The results are shown in Fig. 2.Although FSE can provide outstanding original image and NOE contrast image simultaneously, it takes hours to acquire the images of different frequencies for CEST analysis. The tumor region is obvious in the NOE contrast images from all the three methods, while evident distortion of the whole brain occurs in the result of CEST-EPI because of the naturally inhomogeneous field, which impedes the locating of the tumor area. It should be noted that CEST-SPEN MRI shows a good immunity to the inhomogeneous field and visibly possesses better shapes of brain in both original image and NOE contrast image compared to those of CEST-EPI.The effectiveness of CEST-SPEN MRI and CS reconstruction with edge constraint is demonstrated by experiments on tumor rats. The new CEST imaging scheme proposed here would promote the application of single-shot SPEN MRI.