After computing the FFT on each dSPM file, I smoothed and projected on default. The result is the following structure
This makes sense.
However I have a rather precise hypothesis as for where the effect should appear, so I was trying to use scouts but at this point the option is not available. The only workaround I thought of would be to export the scouts' vertices and proceed through base MATLAB, which seems consistent with your earlier reply. Would you agree?
If you have a precise hypothesis about source localization, the simplest option would be to compute the FFT on the scouts only, and then you don't even have to project on the default anatomy. You can directly compare the values obtained for the various ROIs across subjects.
If you're not afraid about this, you could also script the scout analysis in a Matlab script. This tutorial might help:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting
Note that depending on your sampling rate and epoch duration, the FFT might have a frequency resolution that is too high for having enough overlap between participants.
You could consider using a PSD (https://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsFilter) or frequency bands instead.