Re: [Freesurfer] Fwd: Jacobian mri_cvs_register

Hi Lilla,

Thank you for your response. I would argue that they seem to be aligned, yet they are not. If you look at the cerebellum the Jacobian is shifted in respect to the brain_IBSR_01.mgz. Further, there must be a jacobian for each voxel of the brain_IBSR_01.mgz (at least inside the brain mask), which does not happen (the box is smaller than the brain). Though, I will not discuss this point again.

Just as final note, if anyone want to retrieve the deformation field, a simple way is to virtually create a 4d image that encode the position of each voxel (3 x 3d images), deform this images using the applyMorph, and subtract the initial virtual 4d image. The Jacobian matrix can also be calculated through this fields.

Regards, Andre Santos Ribeiro PhD Student, Imperial College London

On Wed, Aug 6, 2014 at 3:35 PM, Lilla Zollei lzollei@nmr.mgh.harvard.edu wrote:

Hi Andre,

The jacobian field is computed in the coordinate space of the moving volume so if you overlay it with brain_IBSR_01.mgz, you will see that they match.

We currently do not have an executbale in 5.3 that writes out the m3z transforms in nifti format.

Lilla

On Tue, 5 Aug 2014, André Ribeiro wrote:

Hi,

...

The Freesurfer that I am using is: freesurfer-Linux-centos6_x86_ 64-stable-pub-v5.3.0.

The source, target, registered, Jacobian images and deformation field are provided in attachment.

I mean with voxel-wise that by visualizing both (overlayed) there is a shift between the Jacobian and the registered image. I used the same image as target so I would suspect that it is not a header problem (though did not used a different viewer). Additionally, by looking at the Jacobian (and assuming this is correct) the algorithm seems to relaxed allowing really high and low jacobians.

Further, how can I obtain the deformation fields in a nifti format?

​[icon_10_generic_list.png] combined_toIBSR_02_elreg_afteraseg-norm.m3z

​​[icon_10_generic_list.png] brain_IBSR_02.mgz

​​[icon_10_generic_list.png] brain_IBSR_01.mgz

​​[icon_10_generic_list.png] reg_IBSR_01_2_IBSR_02_brain.nii.gz

​​[icon_10_generic_list.png] jacobian.nii.gz

​Regards, Andre Santos Ribeiro PhD Student, Imperial College London

On Tue, Aug 5, 2014 at 4:24 PM, Lilla Zollei lzollei@nmr.mgh.harvard.edu wrote:

  Hi Andre,

  What do you mean by they "do not seem to match voxel-wise"? Are

they of different size or are they not in a corresponding space?

  Which version of FreeSurfer are you using?

  Could you send the volumes over? I don't see anything wrong with

the commands so I would like ot take a look at the files.

  Lilla

  On Tue, 5 Aug 2014, André Ribeiro wrote:

        Hi,

        Ok so most of my problems are now fixed, but there is still

one problem regarding the Jacobian.

        So first of all my images were normalized as follows:
        createMorph --out fullCVSmorph.tm3d --template

f2/mri/brain.mgz --subject f1/mri/brain.mgz --in gcam combined_tof2_elreg_afteraseg-norm.m3z applyMorph --template f2/mri/brain.mgz --transform fullCVSmorph.tm3d vol f1/mri/brain.mgz reg_f1_brain.nii.gz trilinear

        then for the Jacobian:
        mri_jacobian -tm3d combined_tof2_elreg_afteraseg-norm.m3z

f1_brain.nii.gz jacobian.nii.gz

        Now if I visualize both the reg_f1_brain.nii.gz and

jacobian.nii.gz they do not seem to match voxel-wise, and therefore cannot be applied for VBM. Both the jacobian and the registered t1 brain image use the same transform and therefore should be aligned. Am I making some mistake here?

        Sorry for this question in advance.

        Regards,
        Andre Santos Ribeiro
        PhD Student, Imperial College London



        On Fri, Jul 11, 2014 at 4:11 PM, Lilla Zollei <

lzollei@nmr.mgh.harvard.edu> wrote:

              Hi,

                    First, in the mri_cvs_register I kept all the

files and therefore the "combined_toIBSR_02_elreg_afteraseg-norm.tm3d" file was not removed. If I try to replicate this file through the "createMorph --out fullCVSmorph.tm3d --template template.mgz --subject movingvol.mgz --in gcam nlalign-afteraseg-norm.m3z morph el_reg_toTEMPLATE.tm3d" the file produced is different from the one previously created by the program (and gives far worse registration results when used with the "applyMorph"). Is something wrong here?

        No, there is nothing wrong. In the above command you are just

not composing the right files. I assume that you have run the default mri_cvs_register command. In that case there are two non-linear registration steps one aided by aseg info and one by intensities. In the above you are skipping the morph created by the former so that is why you are not getting the same combined output. Instead of "morph el_reg_toTEMPLATE.tm3d" you would need to use "morph combined*aseg.tm3d"

              Second, even if the two files gave the same results

wouldn’t this fullCVSmorph.tm3d (otherwise called combined_toIBSR_02_elreg_afteraseg-norm.tm3d) have both information from *afteraseg-norm.m3z and el_reg_toTEMPLATE.tm3d, and therefore the Jacobian applied solely to the nlalign-afteraseg-norm.m3z would not match.

        It is correct. The jacobians of fullCVSmorph.tm3d and

nlalign-afteraseg-norm.m3z will not match. I am a bit confused about whey you think they should.

              Lastly, am I right assuming the

"combined_toIBSR_02_elreg_afteraseg-norm.tm3d" is the last registration (and most accurate) performed by mri_cvs_register?

        Yes. In the most current FS version though we use the m3z

files as we are trying to phase out the tm3d format.

        Lilla

              On Fri, Jul 11, 2014 at 3:11 PM, Lilla Zollei <

lzollei@nmr.mgh.harvard.edu> wrote:

                    Hi, You should use the final m3z file. The -tm3d

flag just indicates that the file was created by mri_cvs_register. Lilla

                    On Fri, 11 Jul 2014, André Ribeiro wrote:

                          I got an error using the mri_jacobina.
                          The function returns:

                          The input morph originated from a tm3d

(mri_cvs_register file). GCAMread(/home/afs13/Documents/data/test_ freesurfer_reg/combined_toIBSR_02_elreg_aseg.tm3d): invalid version # 0.000

                          mri_jacobian: could not read input morph

/home/afs13/Documents/data/test_freesurfer_reg/combined_ toIBSR_02_elreg_aseg.tm3d

                          The command is provided below:
                          mri_jacobian -tm3d

/home/afs13/Documents/data/test_freesurfer_reg/combined_toIBSR_02_elreg_aseg.tm3d /home/afs13/Documents/data/test_freesurfer_reg/brain.mgz /home/afs13/Documents/data/test_freesurfer_reg/test_jac.nii.gz

                          Regards,
                          Andre Santos Ribeiro
                          PhD Student, Imperial College London



                          On Fri, Jul 11, 2014 at 11:09 AM, André

Ribeiro afs13@imperial.ac.uk wrote: Thank you! Didn't realized that function existed.

                                Regards,
                                Andre Santos Ribeiro

                          On Thu, Jul 10, 2014 at 4:59 PM, Lilla

Zollei lzollei@nmr.mgh.harvard.edu wrote:

                                Hi Andre

                                You can run:

                                mri_jacobian -tm3d <3d morph>

<template volume> <output volume>

                                --Lilla

                                On Thu, 10 Jul 2014, André Ribeiro

wrote:

                                      Dear Freesurfer Community,

                                      How can I obtain the Jacobian

of the deformation field estimated through mri_cvs_register for further use like VBM analysis?

                                      Regards,
                                      Andre Santos Ribeiro



                          ______________________________

---

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

edu/mailman/listinfo/freesurfer

                          The information in this e-mail is intended

only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

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

freesurfer

              The information in this e-mail is intended only for the

person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

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


        The information in this e-mail is intended only for the

person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

---

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.