Computational protocol: fMRI functional connectivity of the periaqueductal gray in PTSD and its dissociative subtype

Similar protocols

Protocol publication

[…] Image preprocessing and statistical analyses were performed using statistical parametric mapping software (SPM12, Wellcome Trust Center for Neuroimaging, London, UK: www.fil.ion.ucl.ac.uk/spm; RRID:SCR_007037) within MATLAB 8.6 (R2015b, Mathworks Inc., MA; RRID:SCR_001622). Four dummy scans were omitted from the fMRI time series to allow magnetization reach steady state before the experiments commenced and enhance the quality of realignment during image preprocessing. The functional images for each subject were realigned to the first functional image to correct for motion in the scanner and resliced. The mean functional image was created and subsequently coregistered to the T1‐weighted structural image for each subject to spatially realign functional images to the subject's anatomical space. The coregistered images were segmented into gray matter, white matter, cerebrum spinal fluid, bone, soft tissue, and air using the “New Segment” method implemented in SPM12, which uses T2‐weighted and PD‐weighted scans when generating tissue probability maps. The resulting forward deformation fields were generated and used to spatially normalize the functional images to MNI space without resampling the voxel size, and each subject was visually inspected to ensure precise normalization patterns given the small anatomical region being studied. The images were then smoothed with a three‐dimensional isotropic Gaussian kernel of 4 mm FWHM (full‐width at half‐maximum), in coordinance with a previous PAG functional neuroimaging study (Dunckley et al., ) and a PAG neuroimaging meta‐analysis (Linnman et al., ) that suggested using a lower smoothing kernel facilitates higher voxel resolution and thus helps elicit optimal functional connectivity patterns based on a smaller neuroanatomical area in the brain (Becerra, Harter, Gonzalez, & Borsook, ). Beissner, Deichmann, and Baudrexel () investigated optimal smoothing and normalization patterns in the brainstem and also found that a relatively lower smoothing kernel may be necessary to obtain significant results given its small region in the brain. It is important to note that this study still satisfies the theory of Gaussian fields developed by Friston et al. (), which recommends that Gaussian smoothing should be a least double the voxel size (2 mm voxel size to 4 mm smoothing).The smoothed functional images were further motion corrected with ART software (version 2015‐10; Gabrieli Lab, McGovern Institute for Brain Research, Cambridge, MA; http://www.nitrc.org/projects/artifact_detect/; RRID:SCR_005994) at a motion threshold of 2 mm, as motion artifacts may significantly affect the BOLD signal in resting‐state functional connectivity studies (Power, Barnes, Snyder Schlaggar, & Petersen, ). The outlier motion regressors identified with ART were used as a covariate of no interest during within‐subject (first‐level) analysis. The smoothed functional images were subsequently bandpass filtered to reduce the signal‐to‐noise ratio using 0.012 and 0.1 Hz as the high‐pass and low‐pass frequency cut‐offs, respectively (in‐house software by coauthor Jean Théberge, Lawson Health Research Institute). [...] A whole‐brain 3 (subject group) × 2 (ROI) full‐factorial analysis of variance (ANOVA) was conducted for the between‐subject analyses, with and without using MDD diagnosis as a covariate (MDD was diagnosed via SCID assessment for Axis‐I psychiatric disorders, see ; Table ). The between‐group factor consisted of three levels: nondissociative PTSD patients (PTSD‐DS), dissociative PTSD patients (PTSD + DS), and healthy controls, whereas the within‐group factor consisted of two levels: DL‐PAG and VL‐PAG. To determine significant clusters, a family‐wise error (FWE) whole‐brain cluster‐corrected (p < .05, k = 50) threshold was set for both interaction and post hoc analyses. One‐sample t‐tests were used to assess connectivity patterns within each group and ROI, whereas two‐sample t‐tests assessed between‐group comparisons for both the DL‐PAG and VL‐PAG as well as the differences between both ROIs. Brain regions were identified using the AAL atlas (Tzourio‐Mazoyer et al., ) via xjview software ( http://www.nitrc.org/projects/xjview) and visually inspected using another anatomical atlas focusing on a dissected brain (Montemurro & Bruni, ). To more accurately distinguish between relevant anatomical areas in close proximity, such as the rolandic operculum and insula, masks of each area were created using PickAtlas software according to the AAL atlas and were inspected to ensure proper identification of brain regions. Brodmann areas of these brain regions were also identified using xjview software and the MNI2Tal atlas available online via the BioImage Suite at Yale University ( http://bioimagesuite.yale.edu/mni2tal/; Lacadie, Fulbright, Constable, & Papademetris, ). […]

Pipeline specifications

Software tools SPM, AAL
Applications Magnetic resonance imaging, Functional magnetic resonance imaging
Organisms Homo sapiens
Diseases Genetic Diseases, Inborn