Dear DCM experts,
I've been using DEM_demo_large_fMRI.m in the toolbox/DEM directory to simulate and understand the spDCM.
Specifically, to understand the difference influence that different network structure has on the algorithm, I made different BOLD signals using different structural connectivity matrices (e.g., 5 nodes, 10 nodes with different topologies).
What I'm wondering is that I see so many false positives (while it captures more or less all the true positives as well) if I run spm_dcm_fmri_csd and visualize the parameters using the following line of code:
spm_plot(ci(DCM.Ep.A(:),DCM.Cp(1:n*n,1:n*n)), hold on
bar(pP.A(:),1/4), hold off
--> pP.A is the structural connectivity matrix that I create to simulate different network topologies
I think this is not it, meaning I assume there exists post-processing of some kind to eliminate the false positives.
Also, I wonder how can I visualize this Ep.A matrix such that I can see the connected graphs.
Would be glad to see some advises!
Best wishes,
|