Introducing CANDIShare Schiz_Bull_2008
Atlases Version 1.0. This is the initial set of atlases
generated from the image and segmentation results released for the
Skitzphrenia Bullitin article.
It is based upon MRI images and
segmentation results of all 4 diagnostic groups (Healthy
(N=29), Schizophrenia Spectrum (N=20), Bipolar Disorder with
(N=19), and Bipolar Disorder without Psychosis (N=35)) from the
Frazier JA, Hodge SM, Breeze JL, Giuliano AJ, Terry JE, Moore
Kennedy DN, Lopez-Larson MP, Caviness VS, Seidman LJ,
Makris N. Diagnostic and sex effects on limbic volumes in
bipolar disorder and schizophrenia. Schizophr Bull. 2008
The release contains a zipped tar file of the CANDI Atlases
directory. Within this directory will be atlases for multiple
studies/publications. In this case, the SchitzBull_2008
subdiretory is found. Since an atlas can be made using many
different methods, we expect to see a different subdirectory for
each different atlasing method. This first release is named
Affine_2_MNI152_2mm as it uses linear, affine transformation (fsl
flirt) to the MNI152_2mm target brain. Within this atlas
subdirectory, we find the subdirectories for the 4 subgropus (HC,
SS, BPDwithPsy & BPDwithoutPsy).
Each of these atlas
subdirectories includes a $DIAG_STRS subdirectory which contains a
probbility file for each of the individual anatomc structures.
These probability files are of the type nii.gz and are in the range
of [0,1] representing the probability of the given structure (XX)
in the diagnostic population. The stucture codes are as provided
in 'seg_fill_values.txt' file to
be found in the Documents section of the CANDIShare NITRC
fill values are correct for automatic interpretation using the
Subcortical' color pallet in fslview and the Freesurfer
addition, the diagnostic directory contains a $DIAG_pMAX.nii.gz
file which, in target (MNI152_2mm) spave has the maximum p value of
any structure for each voxel. Also found is the $DIAG_pSEG.nii.gz
which contains the label value for the sucture of the maximum
probability at each voxel location.
Soon, we will release the detailed processing
script that generated these results.