How to export cortical thickness in atlas ROIs?

Discussions
,
1

Hi - new user here. I want to be able to export cortical thickness values, using atlas ROIs, e.g. the Destrieux. I have successfully imported a Freesurfer directory and can see a thickness result, however, when I export as a Matlab file, I think they are just thickness values for each vertex (251110 x 2 double).
I need a thickness value for each of the ROIs in a standard atlas. Is there a simple way to get this?

Thanks
Michael

2

You can drag and drop the thickness maps to the Process1 box and use the process "Extract > Scouts time series" or "Sources > Downsample to atlas" to extract the values averaged by ROI.
Then you can simply use these these values directly from the Matlab structure (load the corresponding .mat file or right-click > File > Export to Matlab).

Does this help?

3

Thank you for the reply. Unfortunately, I don't see either of those options in the Scout or Sources menu. In the results pane, I can drag down a thickness map into the "Files to process" box.
I'm not quite sure how to use processes - the Extract function for example - is there a button somewhere, or do I do this manually?

Thanks
M

4

I recommend you start by following the introduction tutorials until #23 (#9 explains how to use the pipeline editor):
https://neuroimage.usc.edu/brainstorm/Tutorials

5

Thank you Francois - I have managed to solve this now. I can get out the cortical thickness (CT) values with respect to the atlas scouts. I have a follow-up question, if you don't mind, regarding fsaverage. I can change my default anatomy to fsavg, then for a subject specific set of CT values, I then press Project sources > cortex_15002V, for example. Following the same process as before (extract scout values), does this mean I now have my CT values in fsaverage space?
Sorry to be a pain,
Thanks
M

6

No, this won't work. You would need to create another subject which uses the fsaverage template and then project the source maps from one subject to the other.
But since you are interested in ROI averages, I'm not sure what would be the interest of this, I think it would just add a useless layer of interpolations.

7

Thank you.
For context, I have 3 subject groups and merely interested in differences in CT between them. I want to use a principled way to compare CT, rather than by vertex, thus an atlas makes more sense to me, since different subjects can have different numbers of vertices. The next step is when comparing, to measure everything in fsaverage space, using some sort of nearest-neighbour interp, hence why I want CT values there, rather than just native.

8

If you use a ROI-based approach, you already get rid of the spatial problem. You obtain one value per ROI and you don't have any spatial dimension anymore. You can just compare all the values obtained for one given ROI across all the subjects, not having to know where it is located in space anymore. No need for doing any projection or interpolation of any kind before.

The other possible approach is to project the cortical maps to the a template. You can very easily interpolate your individual thickness maps to the default anatomy of your protocol: https://neuroimage.usc.edu/brainstorm/Tutorials/CoregisterSubjects
Project all your cortical thickness maps to the ICBM152 template, and then compute your averages or tests across groups using these projected maps. You can still compute ROI averages of your group-level averages/statistics if needed.

Make sure you also read the statistics tutorial:
https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics