2.5. Processing of resting state fMRI data

Standard resting state fMRI data preprocessing procedures and first-level analyses were performed using SPM12 (http://www.fil.ion.ucl.ac.uk/spm/) and CONN toolbox (Whitfield-Gabrieli and Nieto-Castanon, 2012), including motion realignment, slice-timing correction and spatial normalization of functional images to the Montreal Neurological Institute (MNI) space, MNI152. The functional images were resampled at 2 mm³, and smoothed using a Gaussian smoothing kernel with 8 mm³ full-width at half maximum (FWHM). Linear regression was used to remove confounding factors of blood-oxygen-level-dependent (BOLD) signal variation, including six head motion parameters, white matter (WM), cerebrospinal fluid (CSF) signals and outlier scans. Outlier scans were detected using Artifact Detection Tools (ART) (included in the CONN toolbox), and were defined as scans with composite motion (combined translational and rotational displacements) greater than 2 mm or 2°, or if the global mean signal was greater than 9 SD. If the percentage of outlier scans was ≥20%, we excluded that subject from further analysis.