Self-adapting multi-peak water-fat reconstruction for the removal of lipid artifacts in chemical exchange saturation transfer (CEST) imaging.

Purpose: Artifacts caused by strong lipid signals pose challenges in body chemical exchange saturation transfer (CEST) imaging. This study aimed to develop an accurate water-fat reconstruction method based on the multi-echo Dixon technique to remove lipid artifacts in CEST imaging. Theory and Methods: It is well known that fat has multiple spectral peaks. Furthermore, RF pulses in CEST preparation saturate each fat peak at different levels, complicating fat modeling. Therefore, a self-adapting multi-peak model (SMPM) is proposed to update relative amplitudes of fat peaks using numerical calculation. With the SMPM-based updating, nonlinear least-squares fitting combined with IDEAL (Iterative Decomposition of water and fat with Echo Asymmetry and Least-squares estimation) algorithms was used for water-fat reconstruction and B-0 mapping. The proposed method was compared with the reported 3-point Dixon method and the fixed multi-peak model in a phantom study using a fat-free Z-spectrum obtained from MR spectroscopy acquisition as the ground truth. This method was also validated by in vivo experiments on human breast. Results: In the phantom experiments, the Z-spectrum from the SMPM-based method agreed well with the fat-free Z-spectrum from CEST-PRESS (point-resolved spectroscopy), validating the effective removal of lipid artifacts, while a decrease or a rise that appeared at -3.5 ppm was observed in the Z-spectrum from the 3-point method and the FMPM-based method, respectively. In the in vivo experiments, no lipid artifacts were observed in the Z-spectrum or the amide CEST map from the SMPM-based method in the fibro-glandular region of the breast with high fat fractions. Conclusion: The SMPM-based method successfully removes lipid artifacts and significantly improves the accuracy of CEST contrast.