Skip to content
Melbourne Children's Regional Infant Brain (M-CRIB) atlas
Branch: master
Clone or download
Bonnie Alexander
Bonnie Alexander merged
Latest commit 3deb84c Mar 29, 2017
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
hard_segmentations first commit Oct 28, 2016
structure_probability_maps first commit Oct 28, 2016
M-CRIB_labels_FreeSurfer_format.txt first commit Oct 28, 2016
M-CRIB_labels_ITK_format.txt added description to ITK labels file Oct 28, 2016
readme.md merged Mar 30, 2017
t1_M-CRIB_template.nii.gz first commit Oct 28, 2016
t2_M-CRIB_template.nii.gz

readme.md

M-CRIB atlas

Update

The individual segmentation images and corresponding individual T1 and T2 images are now available for transfer via signed agreement. If this is of interest, please email deanne.thompson@mcri.edu.au.

Files available here include:

  • T2- and T1-weighted templates
  • Hard segmentation images
  • Structure probability maps
  • Segmentation label files

T2- and T1-weighted templates are based on the 10 M-CRIB participants, constructed with ANTS using nonlinear symmetric diffeomorphic normalisation (SyN) as the transformation model. Further details are available in the paper referred to below.

Thresholded hard segmentation images were constructed by registering the 10 parcellations to the T2-weighted template, and selecting the most frequently occurring label across the sample at each voxel. These segmentations have been thresholded in increments of 0.1 probability. Thus, for an image thresholded at 0.7, the given label in any voxel was present in at least 7 of the 10 participants. Sub-threshold voxels have a value of 0. For (effectively) unthresholded hard segmentation, please use the image labeled '0.1'.

Individual structure probability maps are based on registration of binarised labels for each structure for the 10 participants to the T2-weighted M-CRIB template.

Label files in ITK and FreeSurfer formats list structure indices, names, RGB and alpha values.

Templates and hard segmentations based on rigid and affine registration are also available, purely for the purpose of further illustrating the variability in morphology in the M-CRIB sample. However, these will produce suboptimal labeling, so are not recommended for this purpose. These are available in the current repository under the branch 'linear_atlas_versions'. To access this,

git checkout linear_atlas_versions

Images are in the subfolder linear_M-CRIB_atlas_versions

If available, we may add further M-CRIB atlas data via this page.

Please cite this paper when using and referring to the M-CRIB atlas.

You can’t perform that action at this time.