Dear Jokel,

Data sets that were acquired without experimental manipulation can only be investigated if the DCM used contains stochastic terms (either as part of the neuronal state equations [Friston et al. 2008; Daunizeau et al. 2009] or in terms of stochastic inputs such as white and pink noise in DCM for steady-state responses; Moran et al. 2009). This is currently not the case for the implementation of DCM for fMRI in SPM.

In the absence of experimental inputs, a DCM without such stochastic terms will not predict anything but a flat line. For more details, see pp. 3-4 of the recent "10 simple rules" paper: http://www.ncbi.nlm.nih.gov/pubmed/19914382

Best wishes,
Klaas

Von: Jokel Meyer <[log in to unmask]>
An: [log in to unmask]
Gesendet: Samstag, den 26. Dezember 2009, 10:22:41 Uhr
Betreff: [SPM] DCM for resting state data

Dear SPM-experts!

I am working on a data-set of fMRI scans aquired in resting-state. I proprocessed the images (but did not specifiy & estimate any SPM.mat) and extracted main signal/timecourse (first eigenvariat) from a couple of regions of interest. Now I would be very interested in using DCM to investigate directed influences between different regions and test different models between two groups (patients and controls). However I am not sure about the procedure and whether it is possible at all to use DCM as implemented in SPM in this case as DCM requires an estimated SPM.mat. Can I use my extracted signal/timecourse to directely specify DCMs to then calculated coupling parameters and compare different models on the subject but also on the group level?

Very grateful for any hints & any help!

Best,

Jokel