Hi,
after running the "BrainVisa Freesurfer Pipeline" in the "Freesurfer" toolbox (which completed without errors) i have "lh.aparc.annot.tex" and "lh.r.aparc.annot.tex." files for my subjects. is it possible to make one of these .tex files fit the "*_Lhemi.mesh" produced by Morphologist?
merci,
-joost
aparc 2 BrainVisa
Re: aparc 2 BrainVisa
Hello,
For the time being, this is not possible because when you match a texture and a mesh, many precautions are necessary :
- to make sure that texture and mesh are in the same referentiel
- that the texture can be matched on the mesh, if mesh and texture have the same point number (so one texture for one point of the mesh)
Those conditions are not ok for your case. But you can use meshes which are proposed in the "Brainvisa freesurfer pipeline" .
For instance, by clicking on the eye of a texture, Anatomist will be opened and a correct fusion will be displayed.
I read that you ran the BrainVisa Freesurfer Pipeline without errors, but I think there is an error in the ontogy and a mandatory parameter is not filled out at the step 15.
It's about the "InflatedWhiteCurvText" parameter. In order to resolve the problem, you just need to fill out a value for this parameter. This problem is solved for next release.
Isa
For the time being, this is not possible because when you match a texture and a mesh, many precautions are necessary :
- to make sure that texture and mesh are in the same referentiel
- that the texture can be matched on the mesh, if mesh and texture have the same point number (so one texture for one point of the mesh)
Those conditions are not ok for your case. But you can use meshes which are proposed in the "Brainvisa freesurfer pipeline" .
For instance, by clicking on the eye of a texture, Anatomist will be opened and a correct fusion will be displayed.
I read that you ran the BrainVisa Freesurfer Pipeline without errors, but I think there is an error in the ontogy and a mandatory parameter is not filled out at the step 15.
It's about the "InflatedWhiteCurvText" parameter. In order to resolve the problem, you just need to fill out a value for this parameter. This problem is solved for next release.
Isa
Re: aparc 2 BrainVisa
Hi,
>For the time being, this is not possible because when you match a texture and a mesh, many precautions are necessary :
>- to make sure that texture and mesh are in the same referentiel
>- that the texture can be matched on the mesh, if mesh and texture have the same point number (so one texture for one point of the mesh)
>Those conditions are not ok for your case.
i understand this, hence my question
>But you can use meshes which are proposed in the "Brainvisa freesurfer pipeline" .
sure, but what's their added value? i already have *these* meshes in 'FreeSurfer land' and can get lot's of info from them. i don't need *these* meshes in BrainVisa.
merci,
-joost
>For the time being, this is not possible because when you match a texture and a mesh, many precautions are necessary :
>- to make sure that texture and mesh are in the same referentiel
>- that the texture can be matched on the mesh, if mesh and texture have the same point number (so one texture for one point of the mesh)
>Those conditions are not ok for your case.
i understand this, hence my question
>But you can use meshes which are proposed in the "Brainvisa freesurfer pipeline" .
sure, but what's their added value? i already have *these* meshes in 'FreeSurfer land' and can get lot's of info from them. i don't need *these* meshes in BrainVisa.
merci,
-joost
Re: aparc 2 BrainVisa
hi,
I'm going to see that with the person who developed the toolbox.
Isa
I'm going to see that with the person who developed the toolbox.
Isa
Re: aparc 2 BrainVisa
Hi,
I can confirm that you can't do a fusion between meshes coming from Morphologist and texture data coming from FreeSurfer.
The way to compute them are different, so that those data don't share the same structure. You can't do a fusion between the lh.aparc.annot.tex / lh.r.aparc.annot.tex and _Lhemi.mesh (from Morphologist).
I understand you want to do this, but the result will be wrong. To fit texture data, meshes which are proposed in the "Brainvisa freesurfer pipeline", are exactly what you need. These meshes are computed with FreeSurfer and then convert in GIFTI format to display them with Anatomist (software to visualize and handle volumes, meshes ..... which is automatically driven by BrainVISA by clicking on eyes ).
So the question is the difference between Morphologist and FreeSurfer toolbox:
-> Morphologist is a big pipeline to compute many objects from a T1 MRI. You can compute histogram analysis, grey/white segmentation, grey/white meshes, hemisphere meshes, extraction and recognition of sulci ... then you can perform statistical analysis on sulci.
-> The freesurfer pipeline is a tool to complete the visualization of data which are computed with FreeSurfer. In fact, in freesurfer you can compare a subject to the model but you can't directly compare subjects between them. So with the freesurfer pipeline, data are converted in gifti format and a transformation is extracted from the sphere.reg + ico100_7.mesh (a regular structure) then used to resample freesurfer data in order to compare them.
Now if you are interesting in parcellisation with BrainVISA, I think some processes are available. But I don't know if the user level is advanced or not (if yes, this means that processes can be unstable or under development). If you want more details I can see that.
I hope this can help you ...
Isa
I can confirm that you can't do a fusion between meshes coming from Morphologist and texture data coming from FreeSurfer.
The way to compute them are different, so that those data don't share the same structure. You can't do a fusion between the lh.aparc.annot.tex / lh.r.aparc.annot.tex and _Lhemi.mesh (from Morphologist).
I understand you want to do this, but the result will be wrong. To fit texture data, meshes which are proposed in the "Brainvisa freesurfer pipeline", are exactly what you need. These meshes are computed with FreeSurfer and then convert in GIFTI format to display them with Anatomist (software to visualize and handle volumes, meshes ..... which is automatically driven by BrainVISA by clicking on eyes ).
So the question is the difference between Morphologist and FreeSurfer toolbox:
-> Morphologist is a big pipeline to compute many objects from a T1 MRI. You can compute histogram analysis, grey/white segmentation, grey/white meshes, hemisphere meshes, extraction and recognition of sulci ... then you can perform statistical analysis on sulci.
-> The freesurfer pipeline is a tool to complete the visualization of data which are computed with FreeSurfer. In fact, in freesurfer you can compare a subject to the model but you can't directly compare subjects between them. So with the freesurfer pipeline, data are converted in gifti format and a transformation is extracted from the sphere.reg + ico100_7.mesh (a regular structure) then used to resample freesurfer data in order to compare them.
Now if you are interesting in parcellisation with BrainVISA, I think some processes are available. But I don't know if the user level is advanced or not (if yes, this means that processes can be unstable or under development). If you want more details I can see that.
I hope this can help you ...
Isa
Re: aparc 2 BrainVisa
Hi,
thanks.
>I can confirm that you can't do a fusion between meshes coming from Morphologist and texture data coming from FreeSurfer.
>The way to compute them are different, so that those data don't share the same structure. You can't do a fusion between the lh.aparc.annot.tex / lh.r.aparc.annot.tex >and _Lhemi.mesh (from Morphologist).
like i stated before, i already knew that a simple fusion would be insufficient.
>-> The freesurfer pipeline is a tool to complete the visualization of data which are computed with FreeSurfer. In fact, in freesurfer you can compare a subject to the >model but you can't directly compare subjects between them. So with the freesurfer pipeline, data are converted in gifti format and a transformation is extracted from >the sphere.reg + ico100_7.mesh (a regular structure) then used to resample freesurfer data in order to compare them.
i'm still at a loss what the function of the FreeSurfer pipeline is (or will be)? you write: "a transformation is extracted from >the sphere.reg + ico100_7.mesh (a regular structure) then used to resample freesurfer data in order to compare them." what do you mean here? *what* can be compared in BrainVisa using the gifti meshes that cannot be compared in FreeSurfer? mind that FreeSurfer allows for comparisons of all kind's of data between subjects in native space (although these comparisons do depend to some degree on spherical registration of the surface during reconstruction).
to me, the (only?) added functionality of FreeSurfer imported segmentations is their intersection with the T1 pipeline. ofcourse i know that BrainVisa has cortical segmentation options (under the cortical surface toolbox) but combining sulcal measurements with the gyral aparc (FS) segmentation would have added value. i realize that this isn't easy but just my 2$.
merci,
-joost
thanks.
>I can confirm that you can't do a fusion between meshes coming from Morphologist and texture data coming from FreeSurfer.
>The way to compute them are different, so that those data don't share the same structure. You can't do a fusion between the lh.aparc.annot.tex / lh.r.aparc.annot.tex >and _Lhemi.mesh (from Morphologist).
like i stated before, i already knew that a simple fusion would be insufficient.
>-> The freesurfer pipeline is a tool to complete the visualization of data which are computed with FreeSurfer. In fact, in freesurfer you can compare a subject to the >model but you can't directly compare subjects between them. So with the freesurfer pipeline, data are converted in gifti format and a transformation is extracted from >the sphere.reg + ico100_7.mesh (a regular structure) then used to resample freesurfer data in order to compare them.
i'm still at a loss what the function of the FreeSurfer pipeline is (or will be)? you write: "a transformation is extracted from >the sphere.reg + ico100_7.mesh (a regular structure) then used to resample freesurfer data in order to compare them." what do you mean here? *what* can be compared in BrainVisa using the gifti meshes that cannot be compared in FreeSurfer? mind that FreeSurfer allows for comparisons of all kind's of data between subjects in native space (although these comparisons do depend to some degree on spherical registration of the surface during reconstruction).
to me, the (only?) added functionality of FreeSurfer imported segmentations is their intersection with the T1 pipeline. ofcourse i know that BrainVisa has cortical segmentation options (under the cortical surface toolbox) but combining sulcal measurements with the gyral aparc (FS) segmentation would have added value. i realize that this isn't easy but just my 2$.
merci,
-joost
-
kathleenACE
- Posts: 9
- Joined: Fri Apr 17, 2009 10:28 pm
Re: aparc 2 BrainVisa
I just figured out how to do this. See my reply to your other post:
http://brainvisa.info/forum/viewtopic.p ... 5987#p5987
Let me know if you run into any problems or need clarification.
-Kathleen
http://brainvisa.info/forum/viewtopic.p ... 5987#p5987
Let me know if you run into any problems or need clarification.
-Kathleen
Re: aparc 2 BrainVisa
Hi Joost,
In fact, the Freesurfer coordinate system seems very good, but as far as I know in Freesurfer when you read coordinates on the sphere, you read according to the freesurfer spherical atlas.
When you compute data in freesurfer from subjectA_?hsphere to subjectA_?hsphere.reg, actually the mesh structure is always the same. So if the mesh structure of subjectA_?hsphere have 111695 nodes so the subjectA_?hsphere.reg have too 111695 nodes (the same for subject_B, if subjectB_?hsphere have 112654 nodes, so subjectB_?hsphere.reg have too 112654 nodes). Perhaps we don't know the correct option in freesurfer command lines to work in a different way.
So the brainvisa freesurfer pipeline brings a little subtle distinction to compare the meshes. In order to have a correspondance between the indices and their localisations, a normalisation of the number of nodes and their indexation is performed with an icosphere (the ico100_7.mesh file).
Perhaps the missing step, to understand a little bit more, is that you can compute an average mesh for an analysis group from these new meshes.
Did you try processes in the "AverageBrain" directory proposed in the FreeSurfer toolbox ?
There are 2 other processes :
- "AverageBrain -> 1-Creation of a group of subject"
- "AverageBrain -> 2-Average brain mesh"
The average brain mesh can be used for functional data or other ....
The FreeSufer toolbox is only proposed in BrainVISA since the last release. We work step by step to improve this toolbox. I hope the futur release could propose more stable (under advanced and basic level) and new processes.
Thanks for your feedback, it is very useful for us.
Isa
In fact, the Freesurfer coordinate system seems very good, but as far as I know in Freesurfer when you read coordinates on the sphere, you read according to the freesurfer spherical atlas.
When you compute data in freesurfer from subjectA_?hsphere to subjectA_?hsphere.reg, actually the mesh structure is always the same. So if the mesh structure of subjectA_?hsphere have 111695 nodes so the subjectA_?hsphere.reg have too 111695 nodes (the same for subject_B, if subjectB_?hsphere have 112654 nodes, so subjectB_?hsphere.reg have too 112654 nodes). Perhaps we don't know the correct option in freesurfer command lines to work in a different way.
So the brainvisa freesurfer pipeline brings a little subtle distinction to compare the meshes. In order to have a correspondance between the indices and their localisations, a normalisation of the number of nodes and their indexation is performed with an icosphere (the ico100_7.mesh file).
Perhaps the missing step, to understand a little bit more, is that you can compute an average mesh for an analysis group from these new meshes.
Did you try processes in the "AverageBrain" directory proposed in the FreeSurfer toolbox ?
There are 2 other processes :
- "AverageBrain -> 1-Creation of a group of subject"
- "AverageBrain -> 2-Average brain mesh"
The average brain mesh can be used for functional data or other ....
Thanks for your proposition, you are right it could be an interesting interaction between both software. We have begun this kind of interaction to import principally the grey/white classification computed by FreeSurfer (ribbon file) and to use a part of the Morphologist Pipeline (skipping the grey/white classification of Morphologist and some other steps). If you don't know this process, it is located in "Toolbox FreeSurfer -> Import -> Import From FreeSurfer to T1 pipeline"."to me, the (only?) added functionality of FreeSurfer imported segmentations is their intersection with the T1 pipeline. ofcourse iknow that BrainVisa has cortical segmentation options (under the cortical surface toolbox) but combining sulcal measurements with the gyral aparc (FS) segmentation would have added value. i realize that
this isn't easy but just my 2$."
The FreeSufer toolbox is only proposed in BrainVISA since the last release. We work step by step to improve this toolbox. I hope the futur release could propose more stable (under advanced and basic level) and new processes.
Thanks for your feedback, it is very useful for us.
Isa
Re: aparc 2 BrainVisa
Hi,
>In fact, the Freesurfer coordinate system seems very good, but as far as I know in Freesurfer when you read coordinates on the sphere, you read according to the >freesurfer spherical atlas.
>When you compute data in freesurfer from subjectA_?hsphere to subjectA_?hsphere.reg, actually the mesh structure is always the same. So if the mesh structure of >subjectA_?hsphere have 111695 nodes so the subjectA_?hsphere.reg have too 111695 nodes (the same for subject_B, if subjectB_?hsphere have 112654 nodes, so >subjectB_?hsphere.reg have too 112654 nodes). Perhaps we don't know the correct option in freesurfer command lines to work in a different way.
correct. *native space* meshes in FS, between-subjects, do not have node correspondence. if you want to do vertex-wise group analysis of some metric (e.g. thickness) then meshes of subjects are in some standard space after (spherical) registration and -obviously- have node correspondence. from these aligned meshes you can compute an average mesh. FYI: most of this is accomplished with the 'qcache' command in Freesurfer.
>Thanks for your proposition, you are right it could be an interesting interaction between both software.
what i really needed was the transformation between Freesurfer and Brainvisa, which i have now. this allows me to use the brainvisa generated sulcal medial meshes in freesurfer (e.g. with original -not resampled-not aims- freesurfer meshes and segmentations); this is useful because Freesurfer's sulcal segmentation is less powerful than Brainvisa's. E.g. i can now generate .tmtktri meshes from original freesurfer meshes and use these for cortical metrics which need the sulcal medial meshes and brain surface meshes. i hope this gives you guys some clues about what the added functionality of brainvisa for freesurfer might be.
merci,
-joost
>In fact, the Freesurfer coordinate system seems very good, but as far as I know in Freesurfer when you read coordinates on the sphere, you read according to the >freesurfer spherical atlas.
>When you compute data in freesurfer from subjectA_?hsphere to subjectA_?hsphere.reg, actually the mesh structure is always the same. So if the mesh structure of >subjectA_?hsphere have 111695 nodes so the subjectA_?hsphere.reg have too 111695 nodes (the same for subject_B, if subjectB_?hsphere have 112654 nodes, so >subjectB_?hsphere.reg have too 112654 nodes). Perhaps we don't know the correct option in freesurfer command lines to work in a different way.
correct. *native space* meshes in FS, between-subjects, do not have node correspondence. if you want to do vertex-wise group analysis of some metric (e.g. thickness) then meshes of subjects are in some standard space after (spherical) registration and -obviously- have node correspondence. from these aligned meshes you can compute an average mesh. FYI: most of this is accomplished with the 'qcache' command in Freesurfer.
>Thanks for your proposition, you are right it could be an interesting interaction between both software.
what i really needed was the transformation between Freesurfer and Brainvisa, which i have now. this allows me to use the brainvisa generated sulcal medial meshes in freesurfer (e.g. with original -not resampled-not aims- freesurfer meshes and segmentations); this is useful because Freesurfer's sulcal segmentation is less powerful than Brainvisa's. E.g. i can now generate .tmtktri meshes from original freesurfer meshes and use these for cortical metrics which need the sulcal medial meshes and brain surface meshes. i hope this gives you guys some clues about what the added functionality of brainvisa for freesurfer might be.
merci,
-joost