Image processing and voxel-based morphometry (VBM) analysis

Image preprocessing and statistical analyses were performed with SPM12 (Wellcome Institute of Neurology, University College London, United Kingdom) running under MATLAB R2013b (MathWorks, Natick, MA, United States) using Voxel-based morphometry (VBM) toolbox. Data processing steps referred to the voxel based gray matter asymmetry analysis method proposed in 2015 by Kurth et al. [30]. First, the artifacts, pathological and structural abnormalities of all images were examined. Second, three-dimensional images were segmented into cerebrospinal fluid, white matter and gray matter. A specific group template was created to reduce variability between participants, then the images was normalized into the standard Montreal Neurological Institute (MNI) space using the template. Finally, the modulated gray matter images were spatially smoothed, and isotropic Gaussian smoothing with half height and width of 8 mm was selected for convolution; then resampling to 3 mm isotropic voxels to match the same voxel size as preprocessed fMRI data. Thus, for each participant, there was a gray matter volume (GMV) map, and the GMV values were extracted from the GMV map.

Initial general linear model analysis was used to analyze the structural differences of GMV between different groups. Gaussian random-field method (GRF) was used to correct for Voxel p < 0.01 and Cluster p < 0.05, and the activated brain area was defined as the activated range greater than 25 Voxel.

Additional partial Pearson correlation analysis adjusting for age and gender factors was applied to explore the relationship between the GMV changes and behavior measurement changes that displayed significant differences in the above analyses.