For processing the N x N coherency across subjects, is there a way to do it for each participant?
Not from the interface. The process Coherence NxN has to be called separately for each subject.
If you have many subjects, each one is taking a lot of time and you'd like to run sequentially at night or over the weekend, you could maybe try to script this:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting#Loop_over_subjects
With these N x N ROI coherency maps, how can I restrict my statistical analysis to just one ROI to the remaining ROIs?
If you use non-parametric tests in Brainstorm to test your NxN connectivity matrices between two groups of subjects, it doesn't matter if you have extra data in there. The tests for each pair of ROIs is computed independently from the others. For an NxN matrix, you would have NxN independent tests performed. If you run tests for values that you are not interested in, it does not affect the other tests.
There is just the FDR correction for multiple comparisions that you should probably avoid... I have no idea how this is supposed to work out for a connectivity matrix.
@pantazis?