Hello SPMers,
We are having problems with our DCM analysis in that we are not getting
results.
Below we have listed the steps taken in hopes that the SPM community might
notice where we have made our errors.
Thanks for your time and attention.
Look forward to your comments.
Maria
I set up the 6-Var-Design and specified the same model (M1) for each
control participant as follows:
1. I collect the 6-Var SPM.mat file of the first control subject
2. I enter a name for the model (e.g. M1)
3. I collect VOIs (in this case 3 - LC, Thal, DMPFC) from the 6-Var-folder
of each participant
4. Then I define the input specifications, i.e. I specify the variables
included in the design (I do not include variables of no interest to this
analysis but I include our variable EmDC)
5. I set the timing information to default (3 3 3), same with the Echo time
TE =0.04,
6. I set the model options to default: bilinear modulatory effects, one
state per region, no stochastic effects, no centre input
7. I then specify the intrinsic connections and choose all possible ones (3
are already set, I choose the remaining 6).
8. I then specify the effects of EmDC on specific regions and
connections. As we believe EmDC activates brainstem areas directly, I
chose the LC only here and no effects on connections
9. I repeat this for all control participants
10. I use the spm_dcm_post_hoc command and include all 16 DCM_M1 models
created before (Friston & Penny 2011: This routine searches over all
possible reduced models of a full model (DCM) and uses post hoc model
selection to select the best. Reduced models mean all permutations of free
parameters (parameters with a non-zero prior covariance), where models are
defined in terms of their prior covariance. The full model should be
inverted prior to post hoc optimization. If there are more than 16
free-parameters, this routine will implement a greedy search: This entails
searching over all permutations of the 8 parameters whose removal
(shrinking the prior variance to zero) produces the smallest reduction
(greatest increase) in model evidence.
This procedure is repeated until all 8 parameters are retained in the best
model or there are no more parameters to consider.When several DCMs are
optimized together (as in group studies),
they are checked to ensure the same free parameters have been specified and
the log-evidences are pooled in a fixed effects fashion. The outputs of
this routine are graphics reporting the model reduction (optimization) and
an optimized DCM (for every input DCM) that contains reduced conditional
parameters estimates.)
This gives me 16 DCM_opt_M1 files. I can load these files in Matlab and
review the Ep.A, Ep.B and Ep.C values. Ep.B represents the intrinsic
connections between brain ares. Here it should also be visible which
connections have been removed. Unfortunatly, Matlab tells me they are all
zero for DCM_opt_M1...
|