Anatomist and talairach coordinates

[Olivier Coulon]

Hello there,
I (still) have troubles to see clear in the whole transformation/coordinate business.
As far as I understand, when importing a T1 MRI in Brainvisa, the transformation to Talairach is computed. When displaying the image in Anatomist, the red referential is named as "Talairach-AC/PC-anatomist", but the transformation computed by BV is not read, is it ? In fact, if I load a new referential and add the transformation subject_to_talairach-acpc between the red one and the new one, I can then chose this referential for the viewing window, and when i click I get coordinates that look like Talairach (e.g. 0 0 0 in the middle of AC). Are they ?
So I have the feeling that the red referential is named Talairach by default, but it's not.


Another optional question : by default, a "SPM" referential is loaded and a transformation between the red one and this one is also loaded. This transformation is wrong, isn't it ? (since the red one is not Talairach).

Thanks in advance.

olivier

[riviere]

Hi Olivier,

Your're talking about the development anatomist-3.1 version, right ?

The transformation to our "Talairach-AC/PC" is computed during the segmentation pipeline. The importation only creates a referential, and PrepareSubjects writes AC/PC coordinates, but does not compute a transforlation since the brain needs to be segmented before the transformation can be calculated.

Once it is computed, the transformation is loaded into Anatomist when controled by BrainVisa, but Anatomist alone cannot load it automatically since it doesn't know where the .trm file is.

The status of the "red referential" in anatomist is not completely clear, you're right on this point: it is traditionally used by BV as the common referential to links between several subjects. From this point of view it is actually the "Talairach-AC/PC" referential. But the red ref is also the default referential used for all objects when no other info is available, and in this way it's wrong. We still have to work on it to make it all clear, but it will probably not be in the current version.
The "SPM" referential is the referential of the SPM-MNI normalization template. It allows to get SPM coordinates. The transformation is OK when the red ref is really Talairach-AC/PC. And be careful, this ref is right-handed, contrarily to all other refs used in anatomist, it should be correctly handled now (with respect to 3D rendering, inside/outside definition etc), but there may remain one or two surprising behaviours...

Denis

[Olivier Coulon]

Thanks for your answer Denis.

Your're talking about the development anatomist-3.1 version, right ?

Yes.

The "SPM" referential is the referential of the SPM-MNI normalization template. It allows to get SPM coordinates. The transformation is OK when the red ref is really Talairach-AC/PC. And be careful, this ref is right-handed, contrarily to all other refs used in anatomist, it should be correctly handled now (with respect to 3D rendering, inside/outside definition etc), but there may remain one or two surprising behaviours...

I noticed that in this referential, triangle normals had the wrong orientation...

So, if I want to have access to Talairach, I have to load the subject_to_talairach transformation in Anatomist. Once this is done, I do get proper Talirach coordinates, right ?
Is there a Brainvisa process that does all that ?

And one more : is the same thing possible with the current stable version (I never use it but some people I work with do) ?

Thank you !

Olivier

[riviere]

Hi,

I noticed that in this referential, triangle normals had the wrong orientation...

yes, if you assign a right-handed referential to a mesh that has been built in a left-handed ref, normals will be inverted. But if you display a left-handed mesh which has a transformation to a right-handed one, in a right-handed ref window, it should be OK in recent versions of anatomist 3.1.

Once this is done, I do get proper Talirach coordinates, right ?

yes you should.
Moreover if you load a normalized SPM/Analyze or Nifti image, and use the "apply SPM/Nifti transformation referential change" on it (right-click object menu), you should get the correct transformation to the SPM/MNI template ref.

Is there a Brainvisa process that does all that ?

Not specifically a process, but it's done automatically: BrainVisa systematically loads referentials with objects, and also transformations to the Talairach-AC/PC referential, or to the SPM/MNI ref, when they are available. So you just have to set the view in the referential you want to get coordinates in (AC/PC, or maybe the SMP/MNI ref), and then the view should output the coordinates you want.

is the same thing possible with the current stable version

Brainvisa 3.0 behaves the same way for the Talairach-AC/PC referential (althrough there is less informaton in the referentials window). But it doesn't handle right/left-handed refs transformations correctly, so the SPM/MNI ref is not available in anatomist 3.0. BrainVisa may load a transformation to an "inverted" SPM/MNI ref (with all axes inverted) if one is available.

Denis

[Olivier Coulon]

Thank you very much Denis, it makes a lot of sense now.


Olivier