Hi all,
I am writing to seek help with CSD analyses in both resting and task-evoked DCM.
Our project aims to compare the two paradigms using Bayes Data Comparison modules of SPM. We estimated matrix-A effective connectivity (EC) of 9 nodes from both paradigms using spDCM. In the DCM model specification part, we only specified the matrix-A (full-connection, 81 connections) and didn't set the matrices B and C for both paradigms. However, the level of explained variance (EV) from both paradigms (90%) is quite larger than we expected. Is that normal or accepted?
The second question is slightly technical. Because of the system limit, we have to compile the standalone version of SPM and Matlab Compiler Runtime (MCR) at our high-performance computational clusters. When we used the default standalone and MCR 713 to perform the spDCM, it reported errors. When switching to the MCR v98 and SPM12 7771 version, spDCM was well converged and estimated although there is still a segmentation fault after individual convergence information. We are considering whether there is any preferred version of standalone SPM and MCR for spDCM?
Best regards and all the best,
Shufei Zhang
|