BrainVISA

Is there a scaling of the brain that is done for talairaching? I read that BrainVisa uses the 3 points AC, PC, IH and a bounding box for talairaching, what is this bounding box?

Hi,
Yes, 3 scaling factors are applied along the Talairach referential axes, to make the brain match the bounding box of a "standard Talairach" brain.
Denis

Thanks Denis. So BrainVisa uses a standard talairach base template to do the scaling. I want to recreate/output the talairached brain from BrainVisa to use in my group analysis, I wanted to know where these scaling values are stored. The .trm file only contains the translation and rotation values.

Hi,
The scaling factors are included in the .trm matrix. The 3x3 part is not limited to a rotation, but may include shearing and scaling, it's a full affine transformation matrix. However in most cases the scaling factors are close to 1. so they are maybe not really visible when looking at the .trm.
Denis

Hi,
After running my data throught the analysis pipeline, I have a 4x3 *.trm matrix. I want to apply the transformation matrix to the Raw T1 MRI image to generate talairach T1MRI for each subject. I did the following

/tools/brainvisa/bin/AimsResample -i study2.ima -m TALAIRACH-ACPC.trm -o transformed.ima

But there is only half the brain in the output, given below is my matrix (TALAIRACH-ACPC.trm)
-155.187960 -118.471694 -89.659727
1.270859 0.059891 -0.112749
-0.040245 1.126839 0.144882
0.115250 -0.152473 1.217995

How do I get the whole brain tranformed output?

Hi,

The .trm file seems correct (the command line too). Is it created by the anatomical pipeline in BrainVISA ?
The first line of the .trm file is just coordinates for a translation.
I'm wondering if the problem will be rather with the visualization in Anatomist. It would be a problem with referential of your image or windows (which contains your image).
Have you ever handled referential system in Anatomist ?
Do you use Anatomist to visualize data ? otherwise, which software do you use ?

Isa

Yes, the .trm file is created by the anatomical pipeline in BrainVisa.

The Raw T1 MRI image I am using is 256x256x120 matrix ( or 256x256x144mm). When I played with the translation values in the .trm file (and applied AimsResample) I was able to get a fuller brain (though still looks shifted). I am using AFNI to view the file.

I guess my question is, how do I modify the .trm file to get the AC in the center of the talairach brain, so I can view it in other software as well?
Also what is the bounding box limits set for talairaching? is it the following?
intercommisure - 23mm
front to AC- 70mm
PC to back- 79mm
AC to top - 74mm
AC to left or right - 68mm

Yes your problem is that our Talairach system puts CA at the origine I guess,
leading to negative coordinates for the left of the volume, which can not be handled by our
resampling command. The solution is to add a translation as you did.

Otherwise, if you open the transformation window of Anatomist, which is dealing with
all the referentials and transformations used in Anatomist, you may find a transformation
from our Talairach referential to the MNI space. The underlying trm file is somewhere
in the distribution.

To be honest, I am not sure that our Talairach transformation is such a good choice
for group analysis, except if you design a method that do not depend strongly on
normalization, which is our usual strategy. If you have a look at the recent paper
of Perrot in Medical Image Analysis, you will see what I mean. By the way
Perrot's sulcus recognition method performs a better affine normalization
based on the sulci that must be output somewhere, I do not know exactly where,
as another trm file. This one is in my opinion much more interesting for normalization.