Hi,
Is there a way to convert a connectivity matrix with the structure [Nr x Ntime x Nfreq] to a symmetrical matrix with the structure [ Np x Nm x Nm] where m is the number of the ROIs and p is the number of windows that corresponds to the number of samples for which the connectivity is computed?
Many thanks,
Diana
I'm not sure I understand your question: the matrices do not have the same kind of information represented. The first one represents frequency and some form of continuity in time, the second has a notion of multiple time windows.
Maybe the connectivity tutorial can help you:
https://neuroimage.usc.edu/brainstorm/Tutorials/Connectivity#On_the_hard_drive
I`ll go into a bit more detail. I went through that tutorial but I still need help resolving this issue. At the moment, Iβm trying to import the data from BST to EEGNET for further graph analysis. They require as input a connectivity matrix with the following specifications: βthe matrix must be a square symmetrical matrix with size πΓπΓπ where π is the number of the nodes (it must be compatible with the node position file that must be entered afterward) and π is the number of windows that corresponds to the number of samples for which the connectivity is computed.β
Do you have any idea about how I can obtain it?
Appreciate your help,
Diana
The tutorial section I pointed out explains how to obtain the symmetrical part (m x m) from the compressed storage in the Brainstorm database.
About the windows (p): I guess this makes sense only if you compute a measure for multiple consecutive time segments (either manually of with one of the "Time-resolved" processes in Brainstorm).
What do you do with the frequency dimension? I have no clue.
Maybe just pick one frequency of your choice?