Brain masks

This page has masks that you can download and use for ROIs, small volume/mask-wise correction, and other things.

The examples should ideally have a description of how the mask was created and the code, along with the image files themselves.

Gray Matter Mask

View of gray_matter_mask.img

When correcting for multiple comparisons, it is advisable to use a gray-matter mask that is limited to regions that you would like to test a priori, but also covers all of those areas adequately. A gray-matter mask based on the canonical template can be used, but it is inadequate without manual clean-up to make sure that all relevant areas are included, and other voxels outside brain are not.

There is a gray matter mask in the repository called gray_matter_mask.img

You can apply this mask when viewing or getting results, and when using FDR or FWE to correct statistic images for multiple comparisons. Using it will influence both types of threshold. Its construction is described here. The colors in the image shown here reflect a combination of prior gray matter values from a single subject and the group template, with some manually edited areas. Its primary purpose is to be used as a binary mask, ignoring these values.

Zip file with mask, region object, and code

White matter and CSF masks

White-matter and CSF spaces in standardized MNI format, for use in diagnostics, normalizing/scaling values, and more.

CanlabVisualization toolbox: canonical_white_matter.img canonical_ventricles.img

Display code: imgs = char(which('canonical_white_matter.img'), which('canonical_ventricles.img')) dat = fmri_data(imgs); orthviews(dat, 'overlay', which('avg152T1.nii'))

Code to create:


Mask files and code
VMPFC liberal VMPFC tight Two VMPFC masks, one liberal and one smaller (tighter). Both were created from neurosynth (5.20.13) using the “reverse inference” map for “VMPFC”. The liberal one was thresholded at p < .05 uncorrected, the tight one at p < .0001 uncorrected, after smoothing the Z-map with a 6 mm FWHM kernel and averaging Z-scores across the left and right hemispheres to create a symmetrical map.

Negative Arousal mask

Mask files and code
This pair of masks is based on Neurosynth, and was created using the union of forward- and reverse-inference maps for “arousal,” “negative,” “autonomic,” “aversive,” and “PAG.” Clusters larger than 10 voxels are retained, and many of the input maps are similar to one another. The code is included. The unsmoothed mask has 38 distinct regions.

Roy 2012 TICS term maps

NIFTI images These are reverse-inference fMRI maps from 15 domains, from Roy et al. 2012 TICS. These were combined and subjected to factor analysis in the paper. They are not standard single-term maps, but were constructed from more complex searches through the database, as described in the paper.

Masks based on previous datasets (example: negative IAPS) This code loads contrast images from negative emotion dataset, and creates masks based on thresholded mean image, separately for different clusters.

LC Mask from Keren et al. 2009

Images available from the authors; our copy for internal lab use only.  Code here NIFTI images A Locus Coeruleus MNI-space image defined from a sample of 44 healthy adults (age range: 19-79 years) using high-resolution (in-plane: 0.4 mm x 0.4 mm) T1-weighted Turbo Spin Echo (T1-TSE) MRI. T1 signal is enhanced due to neuromelanin concentration in the LC.

NPS subregion mask: smoothed FDR05 regions

Images are in repository (/3DheadUtility_lite/canlab_canonical_brains/):

 Positive: weights_NSF_FDR05_positive_smoothed_larger_than_10vox.img
 Negative: weights_NSF_FDR05_negative_smoothed_larger_than_10vox.img

The contiguous regions larger than 10 voxels within the FDR (q < .05) corrected NPS map were smoothed with a FWHM 0.5 mm gaussian kernel. Regions contain actual NPS weights in the smoothed regions.

help/core/brain_masks.txt · Last modified: 2016/08/19 18:00 (external edit)
CC Attribution-Noncommercial-Share Alike 4.0 International
Driven by DokuWiki Recent changes RSS feed Valid CSS Valid XHTML 1.0