Hi,
We don't normally do the pre-stats separately, although it will not
effect the stats, but will mean that if you start a new feat session
and only pass it the filtered_func_data then it only has the smoothed
images to choose from when selecting an example functional image
for display. However, more seriously, the registration might not work
as well as this would also only have the smoothed image to work
with and so not perform as accurately. I would recommend doing
everything in the one analysis. You can re-run other contrasts
(post-stats only) at a later date with very little extra computation
and get new results very quickly that way. However, if you want to
run a whole new GLM, with new EVs, then you would need to
run everything again. I would not have expected this to be a
common occurrence, and the pre-stats do not take that long, so
it does not seem like a problem to me.
All the best,
Mark
On 12 Dec 2008, at 05:35, shih-Wei Wu wrote:
> Dear all,
>
> I found that when using filtered_func_data obtained from doing just
> pre-stats (motion
> correction, filtering, and etc) as the input 4D image for Stats and
> Post-stats, the EPI image
> where the zstats was overlayed on looks a bit more blurred compared
> to when just running
> Full analysis. the zstats looks the same.
>
> This is rather trivial, i think, but just wondering why there is
> such a difference.
>
> Also, as a new user, i am just wondering if this is a standard
> procedure where you would
> run prestats first, and then use the filter_func_data for the
> subsequent steps (stats,
> poststats, registration) because every time you want to run a new
> GLM, you don't have to
> run pre-stats again. I imagine that would save some time.
>
> thanks
> shihwei
>
|