16 Apr
2019
16 Apr
'19
7:04 p.m.
A voxel is a 3D cube. The surface is a 2D sheet. The surface may pass through the cube in several ways. It might pass right through middle or just nick one of the corners. When we select whether the voxel is inside or outside of the surface, we have to handle these cases where it is partially in the voxel. --fill-ribbon will require that most of the voxel be within the surface. --annot will determine that the voxel will be inside the surface even if the surface just nicks it.
On 4/16/19 6:58 PM, Caroline Chwiesko wrote:
External Email - Use Caution
Could you explain a little more what you mean with "even if the surface just barely goes through it"? I assume that this is something I would understand when I look in more detail on how the conversion is done from surface to volumetric space?
Carolin
On Tue, Apr 16, 2019 at 3:10 PM Greve, Douglas N.,Ph.D. <DGREVE@mgh.harvard.edu mailto:DGREVE@mgh.harvard.edu> wrote:
I think what is happening is that when you use the --annot option, it will fill in a voxel even if the surface just barely goes through it. With --fill-ribbon, it will only fill in a voxel if that voxel is labeled as ribbon in the ribbon.mgz file. The --fill-ribbon is probably going to be more accurate (which seems to be the case in the images you sent). does that make sense? On 4/16/19 4:54 PM, Caroline Chwiesko wrote: > > External Email - Use Caution > > Hi Douglas, > > Hi Douglas, > I attached 2 images. When I look just at the underlying anatomical > image, I would say it looks like the label voxel covers some white > matter as well. I marked those voxels with two arrows where it is most > easy to see/distinguish the white matter when I switch between > underlay and overlay. In the second image I have included additionally > the ribbon and it shows that those label voxels are not inside the ribbon. > > > > Screen Shot 2019-04-16 at 1.30.01 PM.png > > > > > Screen Shot 2019-04-16 at 1.28.47 PM.png > > On Tue, Apr 16, 2019 at 10:48 AM Greve, Douglas N.,Ph.D. > <DGREVE@mgh.harvard.edu <mailto:DGREVE@mgh.harvard.edu> <mailto:DGREVE@mgh.harvard.edu <mailto:DGREVE@mgh.harvard.edu>>> wrote: > > I think this is probably coming down to partial volume effects. > Can you > look at the underlying image to see whether cortex actually extends > into the brown voxels? > > On 4/15/19 3:54 PM, Caroline Chwiesko wrote: > > > > External Email - Use Caution > > > > Hello Douglas, > > > > that you very much for your quick response! I tried your suggestion > > and it worked! > > > > However, when double checking my results I noticed something > else, I > > would like to ask about. > > > > The labels that I converted into volumetric space using > mri_label2vol > > with --fill-ribbon are nicely overlapping with the ribbon. > > However, the labels, which I converted into volumetric space (and > > which show the holes) using > > mri_label2vol --annot BN_Atlas.annot --subject 1000 --hemi lh --o > > xxxxxx/BN_Atlas.mgz --identity --temp xxxx/brain.mgz --proj frac > 0 1 > > 0.5 don t overlap exactly with the ribbon. > > The latter contain voxels with values, that are 0 in the ribbon > and 0 > > in the labels converted in volumetric space with mri_label2vol. So, > > the labels generated by mri_label2vol with --fill-ribbon and > > mri_label2vol --annot BN_Atlas.annot differ and the difference > is not > > only the presence of holes in the latter. I attached two images for > > illustration. > > > > My question is, why is this the case and which ones are the > "better" > > ones? The ones that overlap with the ribbon? > > > > Thank you very much in advance! > > Carolin > > > > > > Screen Shot 2019-04-12 at 5.29.44 PM.png > > Screen Shot 2019-04-12 at 5.27.55 PM.png > > > > > > > > On Fri, Apr 12, 2019 at 7:48 AM Greve, Douglas N.,Ph.D. > > <DGREVE@mgh.harvard.edu <mailto:DGREVE@mgh.harvard.edu> <mailto:DGREVE@mgh.harvard.edu <mailto:DGREVE@mgh.harvard.edu>> > <mailto:DGREVE@mgh.harvard.edu <mailto:DGREVE@mgh.harvard.edu> <mailto:DGREVE@mgh.harvard.edu <mailto:DGREVE@mgh.harvard.edu>>>> > wrote: > > > > You can just multiply the output by the value of the index you > > want, eg, > > fscalc final.nii.gz -mul 5 -o final.5.nii.gz > > Then give a different number to each label > > > > On 4/11/2019 9:08 PM, Caroline Chwiesko wrote: > >> > >> External Email - Use Caution > >> > >> Hello Freesurfers developers, > >> > >> I am very new to freesurfer. After getting the brainnetome > >> parcellations using freesurfer, we need to convert the > >> BN_Atlas.annot file from surface space into volumetric > space. But > >> the problem we have is that the converted labels have holes > >> inside (see attached image 1). I am aware that this problem has > >> been mentioned on this mailing list. But the suggested solution > >> to use mri_label2vol –seg aparc + aseg.mgz will not work for us > >> as the Brainnetome script does not output such files. We tried > >> two different ways (see below), which did not result in > success. > >> > >> 1.This command successfully brings the labels from surface to > >> volumetric space but gives us the holes inside the labels. > >> > >> Mri_label2vol --annot BN_Atlas.annot --subject > >> 1000 --hemi lh --o xxxx/xxx/xxxxx/xxxxx/BN_Atlas.mgz --identity > >> --temp > >> > >> xxx/xxxxxx/xxxxx/brain.mgz --proj frac 0 1 0.5 > >> > >> > >> 2.I emailed the Brainnetome authors and they suggested the > >> command below. (In the command below I am trying to convert > only > >> 9 Brainnetome labels, just to see, if it will work). > >> > >> > >> mri_annotation2label --subject 1001_20180507_MPRAGE --hemi lh > >> --annotation BN_Atlas --sd > >> /tmp/yassamri/R21/caro/Freesurfer/Freesurfer --labelbase > >> > /tmp/yassamri/R21/caro/Freesurfer/Freesurfer/1001_20180507_MPRAGE/test/lh.BN > >> > >> mri_label2vol --label lh.BN-000.label --label lh.BN-001.label > >> --label lh.BN-003.label --label lh.BN-005.label --label > >> lh.BN-007.label --label lh.BN-009.label --label lh.BN-009.label > >> --label lh.BN-011.label --label lh.BN-013.label --subject > >> 1001_20180507_MPRAGE --sd > >> /tmp/yassamri/R21/caro/Freesurfer/Freesurfer/ --hemi lh > >> --identity --fill-ribbon --o final.nii.gz > >> > >> > >> This approach results in converted labels in volumetric space > >> without holes, but all the labels have the value 1 (see > attached > >> image 2). But we need to keep the correct Brainnetome label > >> values in volumetric space. What command/options could I use to > >> overcome the problem with the holes and convert the file > >> BN_Atlas.annot into volumetric space > >> > >> while keeping the Brainnetome label values? > >> > >> > >> Thank you very much for your help in advance!! > >> > >> Carolin > >> > >> > >> Our freesurfer version: > >> > >> freesurfer-Linux-centos6_x86_64-stable-pub-v6.0.0-2beb96c > >> > >> we run it on Linux, kernel version 4.15.0-46-generic, Ubuntu > >> 18.04.1 LTS > >> > >> > >> image001.png > >> Screen Shot 2019-04-11 at 5.46.19 PM.png > >> > >> > >> > >> _______________________________________________ > >> Freesurfer mailing list > >> Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>>> > >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > > _______________________________________________ > > Freesurfer mailing list > > Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>>> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > > > > _______________________________________________ > > Freesurfer mailing list > > Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> <mailto:Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > _______________________________________________ > Freesurfer mailing list > Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu <mailto: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