Hi Dominic,

It sounds like creating stimulus timing files is a
lot of work, so you would like to avoid that for
subjects that will be thrown out anyway. But also,
you do not want to repeat the pre-processing steps.
Is that correct?

There is nothing wrong with removing 0 TRs in the
tcat block. That is quite common even, since many
scanners automatically throw out pre-steady state
TRs.

Is the concern for repeating preprocessing that it
takes too long? Does that mean you are doing
non-linear registration, or maybe that there are
too many subjects to process?

There are of course many options here. But while
the regression block can be saved for later, that
might be even more work on your part to be sure
that the two phases are well matched.


You could consider:
1. adding an "exit" statement before 3dDeconvolve
2. passing dummy timing files
3. checking the result
4. if good:
a. fix the timing files
b. have the script initialize and continue with 3dDeconvolve


But really, unless preprocessing takes a really
long time, just run it partially once, delete the
results, and run it fully if it looks okay.

There is also the option of running 3dvolreg by
yourself, since that is likely what you are looking
at.

- rick