Thank you, Prof. Gaser. I used the Template_6_IXI555_MNI152.nii file and applied a threshold of 0.1 to make a mask (mid sagittal view attached). Do you think this looks reasonable? The mask from smoothed mwp1 that we get during VBM is of course slightly more dilated than this one. The reason for going for an objective external mask is that we want to do some machine learning using images where the mask has to be independent (or else there might be subtle differences in the mask that get generated in each cross validation).
Regards
PB
On Sunday, May 23, 2021, 2:51:00 AM GMT+5:30, Christian Gaser <[log in to unmask]> wrote:
Dear Patrick,
you could use the Dartel or Shooting template for masking. It's a 4D image, where GM and WM are in channels 1 and 2 and could be thresholded. CSF can be obtained by using brainmask - GM - WM. However, the use of the absolute threshold usually works fine and the value of 0.1 can be increased up to 0.2 or 0.25 depending on your data.
Best,
Christian
On Thu, 13 May 2021 17:05:40 +0000, Patrick <[log in to unmask]> wrote:
>Dear CAT Experts,
>
>Is there a way to create a binary gray matter, white matter, and cerebrospinal fluid masks in CAT12 such that the masks are in DARTEL MNI space? I am interested in a mask that is independent of my dataset (i.e., I don't want to apply a simple threshold to my mwp* images to make a mask). I have looked into the various templates in the CAT folder and the CAT.nii atlas can be useful for making masks for cerebral cortex, cerebellum, etc. However, I did not see any explicit labeling of GM/WM/within brain CSF. Merging various ROIs across atlases is also not an appealing option...
>
>Perhaps an apt analogy will be the "grey", "white", and "csf" files in the OldSeg toolbox or the "TPM" file in the tpm folder of SPM. Are there equivalent version in the DARTEL MNI space? Binarizing the "Template_6_IXI555_MNI152" file is perhaps one way forward, but it does not have CSF and it doesn't seem to have probabilities of tissue classes (rather intensities)...therefore what would be an ideal threshold for each image class?
>
>Any help in this regard is appreciated!
>
>
>Thanks and Regards
>PB
|