Hi Lucy,

The problem here is that when you convert to talairach
space, the resolution difference (from ~4 mm^3 to 1 mm^3)
increases the size of your dataset by a factor of 64. In
such a case, each sub-brick ends up being over 9 Megabytes.
So 840 time points is going to require almost 8 Gigabytes
of disk space, just for a single BRIK file.

In most cases, such a dataset will be too large for your
operating system. So with any software, it is still not
an option (until we can handle it with the 64-bit operating
systems).

One question though, why are you mapping your EPI data to
the +tlrc view? Perhaps is would be more convenient to
just map your statistical data (the post-3dDecovolve data)
to the +tlrc view.

The other option would be to use something like "-dxyz 4"
in your adwarp command. But then your resulting dataset
will match nether the +tlrc grid nor the +orig grid (most
likely).

It would probably be best to run 3dDeconvolve first, then
possibly even extract a few sub-bricks that you care about
from the result, then run adwarp to map that smaller
statistical dataset to Talairach space.

- rick