Dear Sadjad
My apologies for the delay in replying. I hope you don't mind me CC'ing the SPM mailing list so others can benefit.
When you extract the ROIs (with empirical data), any uninteresting regressors in your design matrix are regressed out (i.e. those not included in your effects of interest F-contrast). That includes the mean regressor. Thus, the data will be mean-corrected at this stage. Then, a PCA is applied to summarise the signal across voxels. (All this is implemented in spm_regions.m).
Given you're working in simulation, I guess you start at the point of specifying of the DCM. Is this DCM for fMRI? If so, when you estimate the DCM, the signal is first mean-corrected using the spm_detrend function. Then a scaling is applied to ensure the maximum signal change is 4%. You can find this code between lines 146 and 156 of spm_dcm_estimate.m for fMRI.
I'll have a chat with my colleagues on Monday about working with signals that have meaningful units - there are a few issues to consider. I'll also find out if there's a simple SPM function to use for doing a fourier transform on the inputs - although you could do that directly with Matlab (the timeseries are in DCM.U.u). In the meantime - perhaps you could confirm whether you are using DCM for fMRI or a different DCM. And you mention firing rates - what kind of model are you using to generate the simulated timeseries?
All the best
Peter
-----Original Message-----
From: Sadeghi,Sadjad [mailto:[log in to unmask]]
Sent: 05 July 2017 11:24
To: Zeidman, Peter <[log in to unmask]>
Subject: VOI Time series adjustment
Dear Peter,
I need to compare and match quantitatively the time series of my simulated network (firing rates) with the corresponding predicted time series from DCM analysis (for 3 regions). I want to know if there is any adjustment with the extracted time series used in DCM analysis, are the raw data and predicted data e.g. centered and scaled in some way? If this is the case, how can I transform the firing rates from my simulated network, in the same way, to be comparable? Another question, is there a way to find the frequency of the external inputs that activate the brain regions in an experimental task? I would be so grateful if you answer my questions.
Best Regards,
Sadjad Sadeghi, MSc
Central Institute of Mental Health (ZI)
Department of Theoretical Neuroscience
Laborgebäude, Room 2.10,
J5, 68159, Mannheim,
Germany.
+49 (0)621 1703 2367
|