Hi,
I'm trying to work out how to use a transformation matrix derived from FSL to transform labels from Freesurfer into a standard space (MNI152). If I use mri_label2vol to convert the label to a volume with the scan from subject/mri/orig/001.mgz as a template and then flirt to apply the FSL transformation I end up with something that corresponds correctly to the standard space (and the intermediate volume looks correct compared to the original scan). If on the other hand I multiply the positions in the label file by the transformation matrix from FSL (i.e. M*[x y z 1]') then the labelled region does not appear in the correct space. I have also tried adjusting the translation component of the FSL matrix by half the difference between the standard space and original image dimensions with no success. I've also tried to generate a .mat to pass to mri_label2vol using tkregister2, but despite trying different combinations of I and S in --targ-orientation and --mov-orientation I seem to always have one volume upside down (target if I don't specify the orientation, moveable if I do, the FSL matrix does not have this flip). Any suggestions for things that might be going wrong here or other strategies to try?
Also, assuming I get the transformation right, will adjusting the label point volume to match the new voxel dimensions make any difference if the the threshold is set at zero?
Thanks, Ian