Computational protocol: Lithium alters brain activation in bipolar disorder in a task- and state-dependent manner: an fMRI study

Similar protocols

Protocol publication

[…] Pre-processing and analysis was performed using Statistical Parametric Mapping (SPM), 1999 version []. All functional images were realigned during pre-processing to accommodate and correct for any head motion. Realignment was performed using a 6-parameter rigid body transformation and a mean image was created of the entire time series for each data set. Sessions with realignment parameters of greater than 4 mm in the direction of translation (along the x, y, z axis) were excluded from the final statistical analysis, as were sessions with motion greater than 0.05 radians in a rotational plane (pitch, roll, yaw). The mean image was then spatially normalized to the MNI template brain using a 12-parameter affine transformation with 12 non-linear iterations and 7 × 8 × 7 basis functions. The spatial transformations derived from normalizing the mean image to the template were then applied to the T2* weighted EPI functional images. After normalization, all volumes were resampled to 2 × 2 × 2 mm voxels using tri-linear interpolation in space. Finally, all functional images were smoothed with an 8-mm full width at half-maximum isotropic Gaussian kernel to compensate for between subject variability and allow Gaussian random field theory to give corrected statistical inferences []. Initial analysis was performed separately for each subject for each task. The model specified for each task was kept identical for all subjects and sessions to create identical design matrices. As part of this analysis three more pre-processing steps were performed using SPM99. First the data was high passed filtered to remove low-frequency drifts in the signal. In addition, the data was low pass filtered using the hemodynamic response function to remove high frequency noise. Effects due to global intensity fluctuations were removed when the data was proportionally scaled to a global mean of 100. The time series for each data set was analyzed according to the general linear model. Previously, we have performed a group analysis on 18 volunteers [] by constructing a fixed effects model for each task. Regions of interest for each task were then compiled using the most significantly activated voxels from the group average generation maps and thus pre-determined regions of interested were examined. Region of interest (ROI) images were constructed using automated anatomical labeling software [] running with MRIcro software []. In the present study we assessed activation in these previously identified ROIs.For each subject the individual activation maps generated during single-subject analysis were used to identify the change in BOLD signal magnitude []. The BOLD signal intensity change was calculated based on regions of interest using the MARSBAR toolbox for SPM [] over a seven-voxel sphere centered on the most significantly active voxel in each ROI. The fitted response (or BOLD signal intensity change) is expressed in percentage of whole brain mean. Because the global brain mean in the voxel-wise analysis was scaled to 100, this signal change represents the percentage of signal change with respect to the global mean intensity of the scaled images. The number of voxels over which the fitted response was calculated was kept small in order to minimize averaging over non-significant voxels or large veins []. Subsequently statistical calculations for BOLD signal magnitude were based on the average response calculated from the plateau portion of the hemodynamic response (eight seconds after stimulus origination until stimulus termination. […]

Pipeline specifications

Software tools SPM, AAL, MRIcro
Applications Magnetic resonance imaging, Functional magnetic resonance imaging
Organisms Homo sapiens
Chemicals Lithium