Based on a recent paper looking at simulated and empirical BOLD data
(Sladky et al, 2011, Neuroimage), our study might have the right
combination of TR and trial ITI to see significant benefit from slice
time correction - so we'd like to give it a try.
Our sequence acquired 38 slices with a volume TR of 3.2 seconds.
However, during reconstruction we chucked out the top and bottom 3
slices (by pretending there was oversamping) leaving 32 slices in our
reconstructed images. When doing slice time correction we're faced with
the problem that the timings for acquisition correspond to 38 slices
(i.e. 38 slices, 84.2ms slice tr, 3.2sec volume tr etc), but as far as
spm can 'see' there is only 32 slices in each image. Additionally, our
acquisition is interleaved, from inferior to superior with the even
slices acquired first. So the first acquired slice (slice 2) doesn't
really exist in our images.... our most inferior slice is actually the
second slice that was acquired (slice 4).
Perhaps this means our data aren't suitable for slice time correction?
Any advice on how we could deal with this (and practical help with
implementing that advice - i'm not so hot with code hacking!) would be