DCM experts,
I have motor results from some cortical regions that display a trough in
BOLD followed by a more typical, but much delayed, BOLD peak. Cross
correlation between primary motor and the "odd" regions (contralateral motor
and parietal) shows the delay to be on the order of 21 seconds. My question
is how such a long delay can be dealt with using DCM. While the
time-constant approach of DCM has so far been able to nicely fit the initial
dip intensity, it has not been able to fit lagged rise. I wonder what the
opinions are about whether this is due to the magnitude of the lag (21 sec!)
or to a need for extra terms for fitting a "biphasic" response. If it is
the magnitude of the lag, would this call for inclusion of some sort of
delay (t - delay) into the model in addition to the "decay" constant
(t/tau)? Is there yet another approach that might apply?
Stew Denslow
|