> Suppose that TR=3sec and an event occurs every 6 second.
> The first event occurs when the first slice of the first
> EPI image is acquired.
> If the first slice of each acquisition is used as
> the reference slice, the stimulus onset timing
> should be [0 6 12 18 ...],
This is correct if you enter the onsets in "units of time (secs)".
(Alternatively, you can enter them in units of scans, in which
case they are [0 2 4...]).
> What if the middle slice in time is used as the reference slice?
> Suppose that the number of slices per acquisition is 16
> (TA=[3/16]*15=2.8125)
> and the 8th slice in time is used as the reference slice.
> I know that the value of fMRI_T0 must be changed to 8
> in this case. But I am confused with the SOA specification.
>
> Is it still [0 6 12 18 ...]?
Yes (with fMRI_T0=8).
The onsets do not change. What changes is the specific timebin
(of the 16 per TR) that is chosen for the predicted BOLD signal
for that scan.
> Or is it [-1.3 4.7 10.7 16.7 ...]?
No (though with fMRI_T0=1, this would actually give the same
regressors as above, but we would recommend the above method).
Rik
|