Objective:To investigate the change of chemical exchange saturation transfer (CEST) imaging in the bilateral substantia nigra (SN) and red nucleus (RN) of Parkinson disease (PD), and to explore the value of CEST-MRI for the clinical application of PD.Methods:This was a cross-sectional study. A total of 45 PD patients (PD group) and 21 sex-, age-, and cognitive-function matched normal control subjects (NC group) were retrospectively enrolled from December 2012 to July 2015 in Beijing Hospital. All subjects underwent brain CEST-MRI and routine MRI. Based on the MATLAB software package, the 4-pool Lorentz fitting model was applied to analyze the signal change of CEST imaging, which could acquire the mean amplitudes of the 4-pool parameters including Amide, nuclear overhauser enhancement (NOE), direct water saturation (DS) and magnetization transfer (MT) in the bilateral SN and RN. Independent samples t test and Mann-Whitney U test were used to compare CEST parameters between the PD group and the NC group and controlled by Bonferroni correction. The combined model was constructed based on parameters with inter-group differences after correction. The receiver operating characteristic curve and area under the curve (AUC) were used to evaluate the diagnostic efficiency of the CEST parameters and the combined model. Results:Compared with the NC group, the left SN Amide value, left RN Amide value, and right SN NOE value were reduced in the PD group, and the difference was statistically significant ( t=-3.59, corrected P=0.026; t=-3.77, corrected P=0.016; Z=-3.27, corrected P=0.017). The left SN Amide value, the left RN Amide value, the right SN NOE value, and the combined model all had good diagnostic efficacy in the differentiation of the PD group from the HC group (AUCs of 0.78, 0.79, 0.75, and 0.81, respectively). The combined model had the highest AUC value (0.81) and specificity (97.78%), the Amide value of left SN had the highest sensitivity (93.33%). Conclusions:Quantitative analysis of CEST-MRI based on the 4-pool Lorentz fitting model shows significant differences in the CEST quantitative indicators of the SN and RN between the PD group and the NC group, demonstrating good potential for clinical application in the diagnosis of PD.