All imaging data preprocessing procedures were carried out with Data Processing Assistant for Resting-state fMRI version 4.3,1 which is based on Statistical Parametric Mapping 12.2 The preprocessing procedures include removing first 5 time points, slice-time correction, realignment, co-registration, segmentation for structural images, nuisance covariates regression, normalization to MNI space, and spatial smoothing. The nuisance covariates included the 5 principal components from the individual segmented white matter and the cerebrospinal fluid, 24 motion parameters (6 head motion parameters, 6 head motion parameters one time point before, and the 12 corresponding squared items), and linear and quadratic trends. Particularly, volume-based scrubbing regression by including scrubbing regressors was also included into the multiple linear regression model (52 (link)). The time points with a threshold of framewise displacement (FD) >0.5 mm as well as one back and two forward frames were identified and then modeled as a separate regressor in the regression model of the realigned resting fMRI data. After that, the preprocessed images were temporal filtering. For rsFC, a temporal filtering (0.01–0.1 Hz) was conducted. For effective connectivity, a general linear model (GLM) and an F-contrast analysis were used to identify the low frequency fluctuation in effective connectivity analysis based on previous studies (52 (link), 53 (link)). Specifically, the voxels showing low frequency fluctuations were identified using a GLM containing a discrete cosine basis set with frequencies ranging from 0.0078 to 0.1 Hz. An F-contrast was specified across the discrete cosine transforms, producing an SPM that identified regions exhibiting BOLD fluctuations within the frequency band.
A gray matter mask was generated by including the voxels in which 90% of participants contained EPI signal and the mean gray matter values were larger than 0.2. All of the following analyses were conducted within this mask.
Free full text: Click here