Dear Marko, dear Torben
Dear Torben, dear Cyril,
It is difficult to say in general anything
about this. If for instance there is no motion at all, then the mask
defined by the effects of interest will be very similar in the two
cases. In other cases the difference can be huge.
chiming in on this, I have also played with the piece of code Torben so
kindly provided and tested it in two simple single-session analyses.
There was almost no motion in the first session and quite a bit in the
second one.
In both cases, the p-values were identical in both scenarios
(subtracting spmT[new way] from spmT[old way] is pretty much an empty
image). However, if one checks the effects of interest, the difference
in these spmF-images is rather striking and convincing, removing some
obvious ring artefacts from the effects of interest. So it does do
something.
I also 'played' during the weekend ..
First analyzing another set of data with stronger movements, gave me
different p and T values between the new and old motion correction as
suggested by Torben in the first e-mail (sorry Marko). However, these
differences were not related to the mask (mask new - mask old = 0) ??
Second, the subtraction of different contrasts (con## new - con## old)
gave me either positive or negative results in different parts of the
brain, but most of the time outside regions 'activated'. My guess is
that motion parameters only bias the estimation of regressors of
interest and that's why the (con## new - con## old) computation only
gave 'strong' differences outside activations. However, I still don't
understand the difference in the estimation of these regressors as 'of
interest' or 'of no interest' :'(
. If someone (Torben?) could explain me roughly the difference .. Thank
you in advance.
Additional question: if I have a paradigm with 3 conditions, 2
of interest and 1 where my subjects have to answer .. but I don't want
to look at this .. could I set this condition as 'of no interest'? what
does it means for me when looking at the other two conditions?
Now, I know I am naive about this, but in how
far does this help me if I take my (unchanged) con-images to the second
level? Any advice is appreciated!
Best,
Marko
If you have really no differences in the con images, it is exactly the
same at the second level. Same images.. same results.
Best
Cyril
--
Pernet Cyril
Neuropsychologist, PhD
INSERM U455 Pavillon Riser
CHU Purpan 31059 Toulouse
France
Tel: +33(0)561779503
Fax: +33(0)561499524
[log in to unmask]