analyze to gis conversion and matfile

m.roth · Post by **m.roth** » Tue Aug 14, 2007 9:36 am

Hello everybody,

I want to superpose diffusion images to anatomical image on BrainVISA. I work on analyze images with a matfile. I try the spm to aims converter and then apply a transformation on anatomical image but I obtain as result an image centered on the left corner. Should I have an help on how to use this converter to take into account the matfiles of diffusion images and anatomical image?

Thanks for any help,

Muriel ROTH

Post by **riviere** » Mon Aug 20, 2007 1:42 pm

Hi,

- First, please read http://brainvisa.info/doc/aimsdata-3.0/ ... tials.html

- Are you talking about .mat files with the same name as the image (coregistration information, toto.img, toto.hdr, toto.mat) ? Or normalization information (toto_sn3d.mat/toto_sn.mat) ? I guess they are non-registered.
As far as I remember, the .mat provides a transformation from the image referential (centered on its origin) to "another" referential, this other referential is supposed to be a common exchange referential (two images with a .mat leading to a common referential will be superposed in SPM).
The process in brainvisa extracts the transformation in thei matfile, but combined with the SPM/Aims axes flipping (as far as I remember, too) and origin change. This means it goes from Aims coordinated system of the image (origin in the corner, Aims axes) to the target referential. Here I'm not really sure to remember exactly... but if you combine two of these transformations (in anatomist for instance), it should be OK for superposition.

Denis

m.roth · Post by **m.roth** » Thu Sep 06, 2007 4:07 pm

Hello,

Thank you for your answer. I have already read http://brainvisa.info/doc/aimsdata-3.0/ ... tials.html.
It was useful to understand the difference between the SPM and the aims referentials.

In fact, I just want to take into account the .mat files with the same name as the images. The .mat file provide the transformation between the image referential and the magnet referential. I try to extract and apply this transformation using the brainvisa process. When I observ the resulting image in anatomist, I only see the left upper corner as I told in my first message. Then I try in anatomist to modify the referential (to correct this problem of origin), but I only obtain troncated images. Perhaps it is a problem of bounding box ? Then, it is not really OK for superposition.

Best regards,

Muriel

denghien · Post by **denghien** » Fri Sep 07, 2007 9:36 am

Hi Muriel,

Can you send me data to test them here ?
And can you indicate a maximun of information about them (SPM and Matlab version ...)
Send me a private mail to give you an acces to cea ftp.

Isa

m.roth · Post by **m.roth** » Tue Nov 27, 2007 5:11 pm

Thanks Isa for your help. I found my mistake (I have to change the template image when I apply the transformation).

Best regards,

Muriel