Hi Bob,

We are moving towards only using m3z files and the only reason you still see tm3d files in your output is that you are using the --nocleanup flag so all intermediate results are kept in your output directory. So this means that you will not need to use the applyMorph tool.

The two transformation examples that you gave use transforms from different stages of the registration so they will definitely not produce the same outcome. The final transformation produced by the registration code is final_CVSmorph_toTEMPLATE.m3z. This is the one that you would want to apply to any relevant files.

The transform.txt file is not needed for your purposes. All the information in encoded in the m3z file. When you say that the mri_vol2vol command did not work, I am not sure what went wrong. Can you be more specific? Also, I guess you made a type with the interpolation type, right? "Trillion" is not a valid option.

Lilla

On Thu, 3 Nov 2016, Robert Kraft wrote:

My ultimate goal is apply the CVS transforms to ROIs (similar to aseg.mgz) that are defined in MNI space to anatomical space. Unfortunately, I am having trouble doing this. As a start I am trying to transform my T1w image to MNI space. My understanding so far is that I need to do the following

1. Perform mri_cvs_register to generate the transforms.
2. Then apply the transform using mri_vol2vol

I have been able to perform the mri_cvs_register step with the following commond

nohup mri_cvs_register --mov s001 --mni --nocleanup > nohup.mri_cvs_register &

This worked fine and generated the expected output. The el_reg_tocvs_avg35_inMNI152.mgz overlay with the MNI152 space nicely. As a test of my understand I apply the tm3d transform to the to the original brain.mgz image to see if I can transform the original volume to MNI space. I did this with the applyMorph command

applyMorph --template MNI152_T1_1mm_brain.nii.gz --transform el_reg_tocvs_avg35_inMNI152.tm3d vol ../mri/brain.mgz mni.brain.mgz linear

This worked well and the results are what I expected. Now what I would like to do is do the inverse transform the MNI template into the conformed native space. The applyMorph doesn’t appear to have an inverse capability but mri_vol2vol does. I tried to apply does. In order to understand how the mri_vol2vol command works I transformed the brain.mgz into MNI space with mri_vol2vol with the following command

mri_vol2vol --mov ../mri/brain.mgz --targ MNI152_T1_1mm_brain.nii.gz --m3z combined_tocvs_avg35_inMNI152_elreg_aseg.m3z --o vol2vol_mni_brain.mgz --noDefM3zPath --interp trillion

This didn’t work. It is clear that I am not applying the correct affine transformation matrix first. The mri_cvs_register produces a transform.txt file containing

1.02826 -0.0139273 0.091057 0.0115285 0.990443 -0.222876 -0.0255966 0.234528 0.949605 -3.40075 3.35495 30.3486

However, I am not sure how to incorporate this into the mri_vol2vol function. There are lots of options for the specifying the affine transformation file but the transform.txt doesn’t seem to be in the right format. Frankly, I am just guessing to do from here. I would appreciate any advice or links to documentation. I would also like to know what is the relationship and/or difference between mri_vol2vol and the applyMorph function. Why is there a need for both? When is it appropriate to these functions that seem to do very similar things.

Thanks for your help,

Bob

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer