Problem importing scout atlases

I created and saved a scout atlas about a year ago for one project, but when I tried importing it for a new project, the scout patches are not correct. All that shows up are a series of dots, and they are not in the correct location. Any suggestions for correcting this?
Any example is shown below.

Thanks,
Ann

The scouts are stored as lists of vertex indices, they are valid for one particular surface of one particular subject. The number and order of vertices vary between two cortex surfaces, therefore re-using these lists of vertex indices on a different subject results in what you illustrated.

If the surfaces were all generated using FreeSurfer or BrainVISA, it is possible to use the atlas registration to project scouts from one subject (or one surface) to another. Both subjects must be available in the same protocol, and you should be able to use the menu Scout > Project to > … (from the Scout tab) to project the selected scouts from one subject to another.
http://neuroimage.usc.edu/brainstorm/News#Project_scouts_between_subjects
http://neuroimage.usc.edu/brainstorm/Tutorials/CoregisterSubjects

Thank you for the suggestions. Previously, when I’ve saved and exported custom atlases and then imported them into other projects it hasn’t been a problem. I know now to recreate them for each project.

Thanks!

Hello François. In my case the surfaces were generated with CAT12, and I see on the GUI that it doesn't give me the option to project scouts from one subject to another using the path you mention.

If it helps, I'm basically trying to create a hybrid atlas combining some scouts native to the Desikan-Killiany atlas and some scouts from the Human Motor Atlas Template (HMAT), which I imported manually.

I have the same problem as the user above: I can create the hybrid atlas for one subject, but when I open it on the next subject I get a bunch of dots, due to the vertex indices not matching; as you explained.

So, what's the alternative to projecting across surfaces when the surfaces were generated with CAT12 or spm12? Could you point me in the right direction? Much appreciated.

Surfaces generated with CAT12 include the FreeSurfer registration spheres needed for projecting data and ROIs between subjects.

Scouts on a subject processed with CAT12 v1742:
image

Project to ICBM152 as the default anatomy:


image

Project to the same subject processed with FreeSurfer 7.1:
image

The ROIs might projects as disconnected blobs, because the different brains have different shapes, and there is no one-to-one correspondence between cortex surfaces of different subjects. But overall the shape and coverage is preserved, and it should be OK for source analysis.

A better approach would be to compute your atlases directly at the individual level with CAT (post your question on the SPM mailing list if you are interested).

Note that the Deskian-Killiany scouts should NOT be projected between brains, as they are already available in the subject space for all the subjects processed with CAT and FreeSurfer.

1 Like

Hi François,

Thanks for the information. I'm not sure why, but at first the GUI was not giving me the option to project scouts from one subject kernel to another - or any other kind of projection for that matter - via the scout tab. So, instead, I decided to drop my 5 subjects into Box 1 and project the sources onto the default anatomy via Run > Sources > Project to default anatomy. This took a long time and produced 5 very large group analysis files. But, for some reason, after doing this I now do see the option to project across kernels as shown in your snapshots, regardless of whether I open one of the original surface kernels or one of the massive group-analysis surface files. Can you explain this or point me to a tutorial that would help me understand?

In any case, the good news is that after this procedure I was then able to project the Desikan-HMAT hybrid atlas from one of the original subject kernels to another using the Scout tab path shown in your snapshots. The projected ROIs look good now, but I don’t want to have to project every new subject (or any subject, actually) to the default anatomy template just to get the Scout tab option you say I should have by default. Moreover, you advised me against projecting Desikan scouts across kernels because these scouts are already in subject space, which makes a lot of sense. So, I’m thinking that the best approach would be to automate the scout selection and hybrid atlas generation for each participant via a script that I could add to my pipeline.

Put more formally: I want to generate a script which automatically creates a user atlas containing the same very specific scouts from two standardized atlases, so that I can iterate it through my sample. I could of course do this manually for every participant brain kernel, but I’m dealing with a resting state protocol here with around 200 participants, so I strongly prefer to automate this. I already know how to generate and concatenate .m scripts derived from processes done via Box 1, but this scout selection and atlas generation procedure is all done via the Scout tab, so I’m not sure how to go about generating this script I need.

While we’re on the topic, I also wanted to ask you a smaller question about scout labels and tags, because they’re giving me some trouble when I look at the results of my NxN imaginary coherence connectivity matrix in graph form. As you can see in the ugly image attached, the HMAT scouts are all piled together in the center and the two Desikan scouts are panned to either side based on their “hemisphere tags”. By this I mean the two-letter acronym (for instance: LO = Left Occipital) which can be seen to the left of the scout names in the atlas list under the scout tab itself. My question is: how can I add these “hemisphere tags” to the HMAT atlas scouts, or even remove them from the Desikan atlas scouts in order to resolve this visualization problem? Renaming the scouts via Scout tab > Rename doesn’t work, so I’m out of ideas.

Many thanks in advance,
Oscar

So, instead, I decided to drop my 5 subjects into Box 1 and project the sources onto the default anatomy via Run > Sources > Project to default anatomy. This took a long time and produced 5 very large group analysis files. But, for some reason, after doing this I now do see the option to project across kernels as shown in your snapshots, regardless of whether I open one of the original surface kernels or one of the massive group-analysis surface files.

What you describe does not make me think of any logical explanation. Unfortunately, I don't know what you did precisely and can't tell you why you were not seeing a specific menu. If you can reproduce a precise sequence of processing steps that lead to a behavior you don't understand, please share it in a way that I can reproduce it on my end (list of all the clicks, screen captures, etc).

Can you explain this or point me to a tutorial that would help me understand?

Have you followed all the introduction tutorials using the example dataset provided before trying to process your own recordings?

so I’m not sure how to go about generating this script I need.

I think the only option is for you to edit manually the Atlas structures in the cortex surfaces (be aware not to add in the same atlas two scouts with the same label):
https://neuroimage.usc.edu/brainstorm/Tutorials/Scouts#On_the_hard_drive
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting

My question is: how can I add these “hemisphere tags” to the HMAT atlas scouts, or even remove them from the Desikan atlas scouts in order to resolve this visualization problem?

Set the Region field of the scouts:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scouts#Scout_region
https://neuroimage.usc.edu/brainstorm/Tutorials/Scouts#On_the_hard_drive

Okay I'll make sure to capture all the steps if it happens again. Thank you.

Yes, I'm actually following the OMEGA tutorial very closely, and I refer to the more basic tutorials often. But I'm still a beginner!

I see, that's a shame. Perhaps this is something that could be added to new releases if more people complain about it.

Perfect, thank you. I see that it is very easy to change on the GUI as well.