Surface export not matching MRI space

ArthurBorderie · February 22, 2024, 4:57pm

Hi,

I am working with SEEG data and I am trying to have an overview of a subject in 3D slicer (MRI, TDM, electrods and brain model) in the same space for presentation.

I have imported the electrods coordonates, MRI and TDM from raw data so they are in the same space. However, I am trying to add a brain model created with the pial computed with cat12 in brainstorm and can't find a way to export the pial in the original MRI space.

In the screenshot below you can see to brain models that have been created with stlwrite.m using in yellow pial.fs and in green pial.gii both exported from brainstorm.

Does anyone has a solution ?

Thanks
Arthur

ps: the TDM is in neurology mode (left is right and right is left)

tmedani · February 22, 2024, 6:51pm

Hi @ArthurBorderie

Brainstorm models are in the SCS coordinate system.
Please check this tuto: https://neuroimage.usc.edu/brainstorm/CoordinateSystems#Converting_between_coordinate_systems

To convert the coordinate to the original MRI system, you need to convert first the vertices of your surface and then generate the STL model.
Please check this function that you can use for coordinates conversion:

github.com

brainstorm-tools/brainstorm3/blob/master/toolbox/anatomy/cs_convert.m

function [P, Transf] = cs_convert(sMri, src, dest, P, isNanAllowed)
% CS_CONVERT: Convert 3D points between coordinates systems.
%
% USAGE:       P = cs_convert(sMri, src, dest, P, isNanAllowed=0)
%         Transf = cs_convert(sMri, src, dest)
%
% INPUT: 
%     - sMri         : Brainstorm MRI structure
%     - src          : Current coordinate system {'voxel','mri','scs','mni','acpc','world','captrak'}
%     - dest         : Target coordinate system {'voxel','mri','scs','mni','acpc','world','captrak'}
%     - P            : a Nx3 matrix of point coordinates to convert
%     - isNanAllowed : If 0, throw an error and if the conversion of some points P leads to NaN,
%                      and if there is no linear MNI registration to be used instead
%                      (case of non-linear MNI registration with points outside of the defined area)
%
% DESCRIPTION:   https://neuroimage.usc.edu/brainstorm/CoordinateSystems
%     - voxel : X=left>right,  Y=posterior>anterior,   Z=bottom>top
%               Coordinate of the center of the first voxel at the bottom-left-posterior of the MRI volume: (1,1,1)
%     - mri   : Same as 'voxel' but in millimeters instead of voxels:  mriXYZ = voxelXYZ * Voxsize
%     - scs   : Based on: Nasion, left pre-auricular point (LPA), and right pre-auricular point (RPA).

This file has been truncated. show original

edelaire · February 22, 2024, 7:23pm

hi

the same issue happens when exporting to freesurfer surface (see back here:Exporting surface, not aligned with MRI in FreeView - #2 by tmedani)

ArthurBorderie · February 22, 2024, 8:53pm

Thank you for your quick answer.

The first thing I tried in this project was to use the cs_convert function on the vertices and faces of the pial found in the anatomy folder in the brainstorm database. However, it caused issues with stlwrite.m.
data = load('pial.mat')
Fk = data.faces;
Vk = data.vertices;
sMRI = load('subjectimage.mat')

[Fk, Transf] = cs_convert(sMri, 'scs', 'mri', Fk);
[Vk, Transf] = cs_convert(sMri, 'scs', 'mri', Vk);
stlwrite('test_HEJ_suj12.stl',struct('faces',Fk,'vertices',Vk));
Index in position 2 is invalid. Array indices must be positive integers or logical
values.
Error in stlwrite (line 76)
facets = reshape(facets(:,faces'), 3, 3, );

I supposed that te error was due to the conversion of the surfaces so I tried to apply the .stl to only the vertices to which I applied the cs_convert(scs to mri (i suppose this the correct format)). And As you can see the result is not great to say the least.

Then I tried to export directly the file pial (in brainstom: anatomy>subject>pial>right clic>file>export file>.gii or .fs) thinking that they would be converted back to the original MRI system but as you can see it did not work (1st post).

Is there any other export method that would convert the pial back to the original MRI system ? Or if you have the information of what is the conversion matrix of the exported pial file (from the previous paragraph) ?

ArthurBorderie · February 22, 2024, 8:54pm

Did you find any solution since then ?

edelaire · February 22, 2024, 9:06pm

no sorry