Dear Helmut,
The hemodynamic model which we use for DCM is well capable of modeling the first temporal derivative of the HRF (compare figure 6 in the 2003 DCM paper). Therefore, on theoretical grounds, there should be no problem to apply DCM to a time course that was obtained using a contrast referring to the temporal derivative only. However, if you find your hypothesized activations only for the first temporal derivative, but not for the HRF, I would double check that you have not made any errors in defining your regressors. Often, activations that are specific for the first temporal derivative are indicative of timing problems.
In practical terms, to extract time series for DCM, you should use the “eigenvariate” or “VOI” button. The name changed from the latter to the former at some point in SPM5, I think.
Best wishes,
Klaas
________________________________
Von: Helmut Laufs <[log in to unmask]>
An: [log in to unmask]
Gesendet: Mittwoch, den 12. Mai 2010, 10:27:22 Uhr
Betreff: [SPM] VOI for DCM
Dear DCMers, dear Klaas,
We are about to set up a DCM analysis based on SPM[5]-analyses in which we have modeled the HRF and its temporal derivative. Consistently, we find our hypothesized activated region[s] in the t-contrast for the temporal derivative (but not the HRF). I know there are at least two views on modeling HRF +dHRF/dt and how to assess results (one is to assume the derivative as a confound and just look at the HRF, another to look at F{HRF+TD}, or look at each, the HRF and the derivative etc.).
The two Qs are:
1 (theoretical): As much as DCM is concerned: would it be justified to feed VOI time courses obtained from the dHRF/dt-contrast?
2 (practical): What VOI extraction method should we use (SPM5: what button, what adjustment etc)?
Thanks.
Bw,
Helmut
________________________________
Hotmail: Trusted email with Microsoft’s powerful SPAM protection. Sign up now.
