Integration of multivariate connectivity methods

[robbisg]

Hi,

we are trying to implement a couple of multivariate connectivity methods Multivariate Interaction Measure (MIM) and Multivariate Phase Slope Index (MPSI) to then integrate them with BST.
We created all the files needed such as process_multivariate.m and modified the bst_connectivity to cope with these new metrics.
We are facing some issues in the management of M-scouts since it seems that the scout functions implemented in BST allow only the reduction of the (3x)NxT sources to a 1xT array using first PC or mean or median, are we right?
We would like to obtain the Mx(3xN)xT set of scout timecourses and then apply for example a PCA using the first k-components., in order to run our methods in the M x k x T. Is it possible? How? Thanks a lot
Cheers,
Roberto

[Moo-Marc]

Hello, I'll give a quick answer, but this would be best discussed through the Brainstorm forum. [edit: I mistakenly thought this was a question to the generic bst email and replied by email. While the forum is usually the best place to ask questions, Github is ok too.] Short answer: it's not currently available. You could edit the Brainstorm code to have this option, but it may require some effort. When working on PCA for scouts and for connectivity last year, I tested at some point saving the first 3 components. But we decided to just keep the first. Since there were more development afterwards, I can't easily recover that code to share with you in a usable state. You can have a look at the pcag3 option in this commit to get an idea of how it could be implemented. e3684a3 Hope it helps! Cheers, Marc

[robbisg]

Thanks, I'll post it on the forum! :)

[robbisg]

Just a quick update: we have basically implemented the functionalities, but we need some further tests to then open the PR.