Dear SPM experts, 
we have an experiment with 12 subjects scanned before and after an exercise (PRE and POST). Now we would like to look in specific areas of the brain (i.e. the motor areas) for both activation and deactivation.

We have done the first level analysis and we have obtained *con images for both activation and deactivation, and we have used these images in a second level analysis.
We have obtained a *con image for a) activation PRE  b) activation POST c) deactivation PRE and d) deactivation POST. There seem to be quite a difference. Now the problem is to test whether these differences are significant.


Now according to prior messages in the list:

“I assume you want to use masking as a means of distinguishing "more
activation" from "less deactivation". If so, I think the most convenient
is to do the masking at the stage where the t-maps are displayed.
Let us assume you have taken one con* image per subject to the next
level, and that the subjects are divided into two groups (i.e. a two
sample t-test) which yield two columns in your design matrix.
Furthermore assume you want to know in which areas group 2 activated
more (in the first level contrats corresponding to your con* image) than
group 1. Simply entering the contrast [-1 1] would give you the areas
where group 2 activated more, AND the areas where group 2 deactivated
less. To see only the previous you should mask with the contrast [1 1].”
Date:         Mon, 11 Feb 2002 09:47:55 +0100
Reply-To:     Jesper Andersson <[log in to unmask]>
Sender:       "SPM (Statistical Parametric Mapping)" <[log in to unmask]>
From:         Jesper Andersson <[log in to unmask]>
Subject:      Re: model for group comparisons, and masking
Comments: To: Ulrich Moeller <[log in to unmask]>
Content-type: MULTIPART/ALTERNATIVE;



If we want to compare the PRE vs the POST of both the activation and deactivation if we simply run a paired t-test of PRE vs POST (using the *images of each of the 12 images) PRE > POST will  include what is more active in PRE than in POST, what is non active in PRE and deactivates in POST and what is deactivated in PRE and more deactivated in POST. If we want to separate the activation from deactivation we should use a mask, that is we should mask the 1 -1 contrast in PRE vs POST with a 1 1 contrast to see only what is more active in PRE than in POST and we should mask the 1 -1 contrast of PRE vs POST with -1 -1 to see what is more deactivate in POST than in PRE.

Now the problem is that spm99 and spm2 both won't let me make the 1 1 and the -1 -1 contrasts in the second level analysis (paired t test....)

How can I analyze the differences in deactivation between the two groups???
Please help.

Emanuele