Hi Doug, thanks for the response!
So it looks like my initial transform from surface label to anatomical volume wasn't working because my template.nii.gz and brainmask were in a different space than the functional data I was processing. This happened because I was taking the raw f.nii.gz, doing some transformations first, and saving it to a file called f_reg.nii.gz. I was then trying to run preproc-sess using the -i flag to specify this f_reg.nii.gz, but when preproc runs mktemplate-sess it still uses f.nii.gz to make the template, regardless of what the -i or -mcin flags say. I just added a line in the if($DoTemplate) section of preproc-sess to have the option of adding a -funcstem and it worked fine. I have now updated my preprocessing to run all the commands individually instead of using preproc-sess at all, and I'm probably going to send another message to this list later making sure I'm doing that correctly.
Additionally, Re: "I think you could do it with a single call to mri_label2vol using --regheader and specifying the func template."
This works great! Thanks for the shortcut. I got the surface label direct to functional volume transform working with:
mri_vol2vol --regheader --mov $OUTPUTDIR/rh.MT.volume.anat.nii.gz --targ $MEANFUNCDIR/meanfunc.nii.gz --o $OUTPUTDIR/rh.MT.volume.func-regheader.nii.gz --nearest
But I'm still not entirely sure exactly what information is contained in regheader (or where I can view this information). The reason I have been using regheader instead of register.dof6.lta is that my functional data is aligned to the anatomical prior to running preprocessing, and my understanding was that using regheader would just downsample (similar to mri_convert) and I wouldn't need to specify a registration file. Is this a flawed understanding of regheader?
So transforming to functional space seems to be okay now, but transforming back to the surface is still not perfect.
I tried mri_vol2surf --mov $OUTPUTDIR/rh.MT.volume.func-regheader.nii.gz --regheader $SUBJID --projfrac 0.5 --interp nearest --hemi rh --o $OUTPUTDIR/regheadersurface.nii.gz
and also
bbregister --s $SUBJID --reg $OUTPUTDIR/testreg.lta --mov $MEANFUNCDIR/meanfunc.nii.gz --init-fsl --bold
mri_vol2surf --mov $OUTPUTDIR/rh.MT.volume.func-regheader.nii.gz --reg $OUTPUTDIR/testreg.lta --projfrac 0.5 --interp nearest --hemi rh --o $OUTPUTDIR/testregsurface.nii.gz
Which both give essentially the same result: Original Label: http://web.mit.edu/dsbeeler/www/images/MTlabel.png Back to Surface Overlay: http://web.mit.edu/dsbeeler/www/images/MTpatchy.png This is actually much better than it was, but is there a way to get rid of the patchiness?
Again, thanks for all the useful info!
David
______________________________
Hi David, I've tried to answer your questions below.
doug
On 03/30/2018 01:48 PM, David Beeler wrote:
Hi all, Sorry in advance, this is a bit of an annoying one!
I am trying to use freesurfer's automatically generated MT label as a parcel to constrain activity measured during a functional MT localizer. I am currently doing my analysis in the volume and I'd like to transform the MT surface label into the subject's volume functional space (i.e. the same space as the sig.nii.gz for a particular contrast), take the voxels in sig.nii.gz that are within the MT label, and project them back onto the surface (preferably to be viewed as an overlay with freeview). I will eventually run the analysis on the surface as well (making this much less roundabout, but I also use volume parcels so understanding how to correctly go between the volume and surface is important to me).
In my pipeline I start off by running preproc-sess, then I register and transform my preprocessed functional data (fmcpr.nii.gz) to the subject's anatomical (without upsampling). After doing mkanalysis and mkcontrast, I run selxavg on this volume. So the functional data is aligned with the anatomical, but still has low res dimensions.
If it's not too much to ask, would it be possible to provide the specific commands to do all these transformations properly? I'll provide some of the commands I've tried below so you can see where I'm going wrong:
First I want to transform the MT label to the functional volume. I've tried:
mri_label2vol --subject $SUBJID --label $SUBJECTS_DIR/$SUBJID/label/rh.MT.label --o $OUTPUTDIR/rh.MT.volume.anat.nii.gz --temp $SUBJECTS_DIR/$SUBJID/mri/orig.mgz --identity --fill-ribbon --hemi rh
...which I would expect to put the label in high resolution anatomical volume space (orig.mgz), but it doesn't fill the graymatter ribbon nicely at all.
I would have expected this to work. Can you send a pic? I just ran it myself, and it seemed to do ok.
I then use mri_vol2vol to transform to functional/sig.nii.gz space (I could also just downsample with mri_convert since I've already aligned my sig.nii.gz to my anatomical, this gives me the same result). I am using regheader here, is this okay?
mri_vol2vol --regheader --mov $OUTPUTDIR.rh.MT.volume.anat.nii.gz --targ meanfunc.nii.gz --o rh.MT.volume.func.nii.gz --nearest
...and this works pretty well, but it's a little sparse so I may need to dilate it with mri_morphology. I think this step is fine.
Why do this extra step? I think you could do it with a single call to mri_label2vol using --regheader and specifying the func template. Why are you using --regheader here instead of the register.dof6.lta?
Next I use matlab to take the voxels in sig.nii.gz that are also in rh.MT.volume.func.nii.gz and I save that as a new volume, say constrainedsig.nii.gz.
You can do this with mri_mask sig.nii.gz rh.MT.volume.func.nii.gz constrainedsig.nii.gz
Now I want to transform constrainedsig.nii.gz to the surface. I have tried going directly using the register.dof6.lta generated from preproc-sess:
mri_vol2surf --mov constrainedsig.nii.gz --reg register.dof6.lta --projfrac 0.5 --interp nearest --hemi rh --o $OUTPUTDIR/constrainedsig_surface.nii.gz
This should have worked. Try doing directly on the rh.MT.volume.func.nii.gz
... and this is in the right area, but its shape doesn't match the shape of the original label. So I've also tried backtracking and doing another vol2vol and then vol2surf with regheader:
mri_vol2vol --regheader --mov constrainedsig.nii.gz --targ $SUBJECTS_DIR/$SUBJID/mri/orig.mgz --o $OUTPUTDIR/constainedsig_anatvolume.nii.gz
mri_vol2surf --mov constrainedsig_anatvolume.nii.gz --hemi rh --regheader S02 --o $OUTPUTDIR/constrainedsig_anatsurface
...and while this ends up in the right place, it is really patchy (looking kind of like twisty zebra stripes). Seems to me like it has the wrong dimensions or something, but mri_info says it's the same dimensions as the volume I originally created with mri_label2vol.
Any help is much appreciated, thanks! David
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
On 04/11/2018 12:44 PM, David Beeler wrote:
Hi Doug, thanks for the response! So it looks like my initial transform from surface label to anatomical volume wasn't working because my template.nii.gz and brainmask were in a different space than the functional data I was processing. This happened because I was taking the raw f.nii.gz, doing some transformations first, and saving it to a file called f_reg.nii.gz. I was then trying to run preproc-sess using the -i flag to specify this f_reg.nii.gz, but when preproc runs mktemplate-sess it still uses f.nii.gz to make the template, regardless of what the -i or -mcin flags say. I just added a line in the if($DoTemplate) section of preproc-sess to have the option of adding a -funcstem and it worked fine. I have now updated my preprocessing to run all the commands individually instead of using preproc-sess at all, and I'm probably going to send another message to this list later making sure I'm doing that correctly. Additionally, Re: "I think you could do it with a single call to mri_label2vol using --regheader and specifying the func template." This works great! Thanks for the shortcut. I got the surface label direct to functional volume transform working with: mri_vol2vol --regheader --mov $OUTPUTDIR/rh.MT.volume.anat.nii.gz --targ $MEANFUNCDIR/meanfunc.nii.gz --o $OUTPUTDIR/rh.MT.volume.func-regheader.nii.gz --nearest But I'm still not entirely sure exactly what information is contained in regheader (or where I can view this information). The reason I have been using regheader instead of register.dof6.lta is that my functional data is aligned to the anatomical prior to running preprocessing, and my understanding was that using regheader would just downsample (similar to mri_convert) and I wouldn't need to specify a registration file. Is this a flawed understanding of regheader?
Regheader assumes that the two volumes "share a RAS space". This is a little hard to explain. Each volume has some geometry information in it that determines where a voxel is in RAS (right, anterior, superior) space. --regheader assumes that the RAS space is the same, so, if the RAS for voxel in volume 1 is the samseg as that of a voxel in volume2, then they must represent the same place anatomically. If you have already resampled your functional data into the 1mm, 256^3 anatomical space, then --regheader should work ok
So transforming to functional space seems to be okay now, but transforming back to the surface is still not perfect. I tried mri_vol2surf --mov $OUTPUTDIR/rh.MT.volume.func-regheader.nii.gz --regheader $SUBJID --projfrac 0.5 --interp nearest --hemi rh --o $OUTPUTDIR/regheadersurface.nii.gz and also bbregister --s $SUBJID --reg $OUTPUTDIR/testreg.lta --mov $MEANFUNCDIR/meanfunc.nii.gz --init-fsl --bold mri_vol2surf --mov $OUTPUTDIR/rh.MT.volume.func-regheader.nii.gz --reg $OUTPUTDIR/testreg.lta --projfrac 0.5 --interp nearest --hemi rh --o $OUTPUTDIR/testregsurface.nii.gz Which both give essentially the same result: Original Label: http://web.mit.edu/dsbeeler/www/images/MTlabel.png Back to Surface Overlay: http://web.mit.edu/dsbeeler/www/images/MTpatchy.png This is actually much better than it was, but is there a way to get rid of the patchiness?
The patchiness is caused by something else. When you run vol2surf, you can try --projfrac-max -.1 1 .1 That should reduce the patchiness. If it is still not tolerable, then let me know
Again, thanks for all the useful info! David ______________________________ Hi David, I've tried to answer your questions below.
doug On 03/30/2018 01:48 PM, David Beeler wrote:
Hi all, Sorry in advance, this is a bit of an annoying one!
I am trying to use freesurfer's automatically generated MT label as a parcel to constrain activity measured during a functional MT localizer. I am currently doing my analysis in the volume and I'd like to transform the MT surface label into the subject's volume functional space (i.e. the same space as the sig.nii.gz for a particular contrast), take the voxels in sig.nii.gz that are within the MT label, and project them back onto the surface (preferably to be viewed as an overlay with freeview). I will eventually run the analysis on the surface as well (making this much less roundabout, but I also use volume parcels so understanding how to correctly go between the volume and surface is important to me).
In my pipeline I start off by running preproc-sess, then I register and transform my preprocessed functional data (fmcpr.nii.gz) to the subject's anatomical (without upsampling). After doing mkanalysis and mkcontrast, I run selxavg on this volume. So the functional data is aligned with the anatomical, but still has low res dimensions.
If it's not too much to ask, would it be possible to provide the specific commands to do all these transformations properly? I'll provide some of the commands I've tried below so you can see where I'm going wrong:
First I want to transform the MT label to the functional volume. I've tried:
mri_label2vol --subject $SUBJID --label $SUBJECTS_DIR/$SUBJID/label/rh.MT.label --o $OUTPUTDIR/rh.MT.volume.anat.nii.gz --temp $SUBJECTS_DIR/$SUBJID/mri/orig.mgz --identity --fill-ribbon --hemi rh
...which I would expect to put the label in high resolution anatomical volume space (orig.mgz), but it doesn't fill the graymatter ribbon nicely at all.
I would have expected this to work. Can you send a pic? I just ran it myself, and it seemed to do ok.
I then use mri_vol2vol to transform to functional/sig.nii.gz space (I could also just downsample with mri_convert since I've already aligned my sig.nii.gz to my anatomical, this gives me the same result). I am using regheader here, is this okay?
mri_vol2vol --regheader --mov $OUTPUTDIR.rh.MT.volume.anat.nii.gz --targ meanfunc.nii.gz --o rh.MT.volume.func.nii.gz --nearest
...and this works pretty well, but it's a little sparse so I may need to dilate it with mri_morphology. I think this step is fine.
Why do this extra step? I think you could do it with a single call to mri_label2vol using --regheader and specifying the func template. Why are you using --regheader here instead of the register.dof6.lta?
Next I use matlab to take the voxels in sig.nii.gz that are also in rh.MT.volume.func.nii.gz and I save that as a new volume, say constrainedsig.nii.gz.
You can do this with mri_mask sig.nii.gz rh.MT.volume.func.nii.gz constrainedsig.nii.gz
Now I want to transform constrainedsig.nii.gz to the surface. I have tried going directly using the register.dof6.lta generated from preproc-sess:
mri_vol2surf --mov constrainedsig.nii.gz --reg register.dof6.lta --projfrac 0.5 --interp nearest --hemi rh --o $OUTPUTDIR/constrainedsig_surface.nii.gz
This should have worked. Try doing directly on the rh.MT.volume.func.nii.gz
... and this is in the right area, but its shape doesn't match the shape of the original label. So I've also tried backtracking and doing another vol2vol and then vol2surf with regheader:
mri_vol2vol --regheader --mov constrainedsig.nii.gz --targ $SUBJECTS_DIR/$SUBJID/mri/orig.mgz --o $OUTPUTDIR/constainedsig_anatvolume.nii.gz
mri_vol2surf --mov constrainedsig_anatvolume.nii.gz --hemi rh --regheader S02 --o $OUTPUTDIR/constrainedsig_anatsurface
...and while this ends up in the right place, it is really patchy (looking kind of like twisty zebra stripes). Seems to me like it has the wrong dimensions or something, but mri_info says it's the same dimensions as the volume I originally created with mri_label2vol.
Any help is much appreciated, thanks! David
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
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
freesurfer@nmr.mgh.harvard.edu
-
David Beeler -
Douglas N. Greve