Coordinates extraction of statistical activation in SPM

Discussions
1

Hi experts,

I have a question regarding SPM-Brainstorm.

I have performed some statistical comparisons in SPM with the MEG source images extracted from Brainstorm. Then I have identified the maximum statistical significance voxels’ coordinates (in millimeters).
The next step I plan to do is to study functional connectivity in between the closest sources to these voxel’s coordinates. For that I have first extracted the 15002 source time series from Brainstorm to matlab. Then I have loaded the tess_cortex_pial_low.mat. This .mat file has a mesh of 15002 vertices, which I thought, should correspond with the 15002 sources I have previously reconstructed.
Should the coordinates of these vertices match the SPM coordinates? I have been checking the extreme (maximum and minimum) coordinates and I am not sure about that.

Thanks a lot

Gema

2

Hi Gema,

I don’t know how SPM works, I don’t know how to relate the coordinates your read in SPM to anything in Brainstorm.
About the statistics you have run in SPM: was this done on surfaces or re-interpolated in volumes?

If you do your analysis based on the surfaces (SPM12 only), you don’t have to worry about the coordinates, just use the vertex indices.
If you have done your analysis based on volumes, you have to estimate the translation that there could be between the Brainstorm and SPM coordinates.
Take a precise point (for instance AC), check its coordinates in both software (MRI coordinates in Brainstorm), this will give you the translation between the two “MRI/voxel coordinates” in millimeters.

Then, you need to convert again between the MRI and the SCS coordinates in Brainstorm (the coordinates of the vertices are in SCS coordinates).
This page will help you with this: http://neuroimage.usc.edu/brainstorm/CoordinateSystems

Cheers,
Francois

3

Hi Francois,

thanks for your answer. I understand you’re not expert in SPM, so I will also post this issue in the SPM list. But any advice will be of great help.

Im using SPM8, and my analyses are based in volumes. I tried as you said, to find a linear transformation betwen AC coordinates, however it was not necessary, as both AC in SPM and Brainstorm had the same MNI coordinates.

In SPM, when I visualize the MNI template, the coordinates match exactly with those of MNI in Brainstorm, when I visualize the Collins template.
However, when I export my sources in matlab, the vertices coordinates do not match these of MNI. These coordinates range in the following intervals [X(-83,97),Y(-71,72),Z(-7,131)]. I think that these values are also in the standard space, as they correspond to the Collins template, but I need to transform volume coordinates to them. Is this transformation stored in any file?

Thanks a lot.

Cheers
Gema

4

Hi Gema,

The coordinates you read that are similar in Brainstorm and SPM are "MNI coordinates".
The coordinates in which the surface vertices are saved are in "Subject Coordinate System" (= SCS = CTF coordinates).
You need to convert from "MNI coordinates" to "MRI coordinates", and then from "MRI coordinates" to "SCS coordinates".

The page I indicated previously shows how to perform those conversions:
http://neuroimage.usc.edu/brainstorm/CoordinateSystems#Converting_between_coordinate_systems

P_scs = surface.Vertices;
P_mri = cs_scs2mri(sMri, P_scs' .* 1000)';
P_mni = cs_mri2mni(sMri, P_mri' .* 1000)' ./ 1000;

Francois

5

Hi Francois,

Now its solved.
Thank you very much.

Gema