Dear Robin,
> Many thanks for the detailed explanation. Yes I think what you said corresponds to what I thought. So when entering microtime onset, we
> should input not the spatial position of the reference slice, but it's temporal position. In other words,for those of us who almost always use
> the temporally middle slice as the reference slice, the microtime onset should always be half of microtime resolution (Eg 8 and 16, or 19/20 and
> 39), regardless of the slice acquisition sequence that was used i.e. Doesn't matter whether the temporally middle slice is no. 20 or 39 of the
> total 39 slices.
Exactly. The spatial position is only relevant during STC, as SPM can only infer the spatial properties of the slices within a volume. Accordingly, we have to define the temporal acquisition scheme by entering the "spatial numbers" of the slices in a certain order.
> A related question: is setting the temporally middle slice as the reference in slice timing correction dependent on the task stimulus onset
> being locked to the onset of TR, as many event-related paradigms are not e.g. due to variable RT and jittered ITI across trials?
No. The idea is to minimise the required temporal shift overall, as it would be +/- TR/2 at max. for some slices when going with the temporally middle slice, but larger (up to a full TR) when going with the temporally first or last slice.
Going with another reference slice might make sense in case you're interested in a particular slice (that contains a certain anatomical region) for which you want to keep interpolations at a minimum. This would best combine with an ascending or descending acquisition scheme, as a single slice might not be sufficient to cover the region entirely. With those acquisition schemes the temporal acquisition gap between spatially neighbouring slices would still be rather small, so the necessary correction would also be limited (in contrast to interleaved schemes).
Best
Helmut
|