Hi,
I have an image in native space, which has values only in a grey matter mask , and NaNs in the rest of the image (It's actually the output of an RSA analysis I've run on the surface with using the Freesurfer reconstruction). I want to move this image to MNI space to do group stats. If I use applywarp, I get many more voxels with NaNs in the output image than in the original image. I assume that's because for every voxel which has a NaN in the original image somewhere in it's vicinity, applywarp will use that NaN to interpolate. So the only voxels that survive are voxels in the middle of the grey matter, which are "padded" with other grey matter voxels, so they have no NaNs when interpolating.
This is an issue because I end up getting one image per subject in MNI space with thin and non-matching NaN patterns, and I can't do group stats because most voxels in MNI space have at least one subject (usually many more with a NaN.
I can't replace the NaNs in native space with zeros, because that would bias the transformed images toward 0 in voxels that are close to NaN voxels (close to the edge of the grey matter). (My images are of "1 minus correlation" values, i.e values between 0 and 2, so pushing voxels towards 0 has a huge effect).
Is there a way to ask applywarp to ignore voxels that are outside a mask when interpolating? or any ideas how to solve this problem?
Thanks very much!
Best wishes,
Alon
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|