Dear FreeSurfer experts,
I am working on some
monkey data and trying to get an accurate nonlinear
transform to talairach space (talairach.m3z in the -careg
stage). The problem is that the monkey brains look so
different from the human brain that this registration ends
up being very far off, even with an accurate talairach.lta.
As an alternative, I did
a two-stage nonlinear registration using ANTs (first from
subject to a macaque template and then from macaque template
to talairach), which gives an accurate nonlinear transform.
However, I can’t seem to convert this nonlinear transform
file correctly from .nii.gz to the .m3z format that
FreeSurfer needs to continue down the pipeline.
How would one go about
converting the nonlinear transform from ANTs to a correct
.m3z file?
Some other things I
tried:
-
I have tried a workaround in which I
applied the inverse ANTs transform to the first frame of
RB_all_2016-05-10.vc700.gca to get the talairach volume in
subject space, and then ran mri_ca_register on that volume
to obtain a talairach.m3z. That transform works but is
still not accurate enough unfortunately.
-
The nonlinear transform itself seems
correct because when I apply it to the movable volume
using ANTs the results are similar to:
mri_convert
movable.mgz –apply_transform talairach.m3z -oc 0 0 0
movable_in_talairach.mgz
where movable.mgz is the talairach volume in
subject space and talairach.m3z is the transform made by
mri_ca_register.
The dimensions, orientation, and voxel size of
movable_in_talairach.mgz and my own transformed volume are
identical.
Any help would be
greatly appreciated!
Platform: MacOS Mojave
Version: FreeSurfer
6.0.0