Hello, I've become confused about the direction of the transformation specified in the talairach.xfm file.
It seems that if you "manually" specify the --targ and --mov volumes in conjunction with the --xfm option, that the "target" must be the talairach average to get proper alignment: e.g., tkregister2 --targ $FREESURFER_HOME/subjects/fsaverage/mri/T1.mgz --mov orig.mgz --xfm transforms/talairach.xfm --reg tmp.dat
(The same targ/mov definition appears to be the case if you want to use --lta to view a .lta transform instead).
However, the resulting output to the terminal for "Input registration matrix" does not match the talairach.xfm file -- is there a matrix inversion somewhere??
But, if you use the --fstal option e.g., tkregister2 --s 020819_43 --fstal
then orig.mgz is now the "target" and the talairach average is now the "movable" volume, and the output to the terminal for "Input registration matrix" now exactly matches the values in the talairach.xfm file. Based on this, it seems that .xfm maps from mov to target (with orig.mgz being the target). However that doesn't seem consistent with the required mov and targ specifications with using the --xfm flag directly...
thanks for clarifying, Mike H.