Greetings AFNI minds,

For our recent experiment, we used a sequential scanning montage, 24 Axial slices collected Inferior to Superior, with a 2 s TR. This was intended to minimize timeshift interpolation requirements to "synchronize" signal data from slices collected from adjacent tissue. We figured 3dTshift would only have to make an 83 ms adjustment for physically adjacent sequential slices versus roughly 1000 ms adjustment for physically adjacent but interleaved slices.

In to3d, I specified the "seq+z" operator to create the dataset. When I subsequently ran 3dTshift on the dataset, did the algorithm align the uppermost (superior) 23 slices to the initial bottom (inferior) slice?

By this I mean that were I to apply a canonical hemodynamic response function to the entire volume+time dataset, AFNI will assume data from the topmost slices of the brain was measured/collected at the same time as the bottom slices, when in reality, the same event time-linked to data obtained from the first slice(s) actually occurred almost 2 s earlier relative to the signal data obtained from the uppermost slices?

It seems to me that if this is the case, the best way to deal with this acquisition montage in terms of modeling signal data to events is to artificially "move up" the timing of events along an 83ms gradient (using -tstim) when modeling data collected in slices collected last (superior) relative to timing of events fed into AFNI for the inferior slice first collected in each pass.

What say?

Jim B