 |
 |
Dear Martin,
Thank you for your answer!
My concern is that I have already used your suggested method and I didn’t find any brain regions with significant activation differences. If I used to define the pre image as -1 and the post image as 1 and build the contrast [-1 1], then the SPM treats both conditions as two independent variables and this is an independent or two-sample t-test. This t- test is less sensitive for treatment effects compared with the paired t-test, because the paired t-test contains less error variance based on the explained inner-subject variance. On the other hand I don’t have a directional hypothesis, thus I have to use two-tailed test.
Thus, I would like to find a programming method (may be with MATLAB or with some kind of SPM-toolbox) allowing me to compute for example the amounts of the beta-weight differences after the paired t-test and so to show on the glass brain one time just the brain regions with negative differences and respectively another time the brain regions with positive differences.
I hope my problem has become clearer now!
Kind regards
Von: Martin Dietz [mailto:[log in to unmask]]
Gesendet: Mittwoch, 3. Februar 2016 17:02
An: Andreas Jell Heinrich-Heine-Universität Düsseldorf
Cc: [log in to unmask]
Betreff: Re: [SPM] Implemting a specific function
Dear Andreas,
The paired t-test in SPM will give you exactly what you want as a function of the contrast you use to implement your hypothesis. If the first image in a pair is your pre and the second image is your post condition, the contrast [-1 1] will test for a greater effect post compared to pre and vice versa, using a one-tailed p-threshold at the positive tail of the null t distribution.
An alternative way of implementing your question is to create contrast images at the first level of the difference between your two conditions using the contrast above and perform a one-sample t-test at the second level with a 1 or -1 contrast, respectively.
I hope this helps
Martin
On 03 Feb 2016, at 16:32, Andreas Jell Heinrich-Heine-Universität Düsseldorf <[log in to unmask]> wrote:
Dear SPM community,
I have a problem with a specific graphic representation of contrast images in the SPM glass brain. The contrast images represent the differences between pre- and post- conditions after the second level analysis by a paired t–test. Because in SPM the paired t–test is two-tailed, both directions of differences are considered and therefore those brain areas involving increased and decreased brain activations are displayed in the SPM glass brain. I would like to separate both directions graphically and to show after the paired t–test analysis one time only brain regions, where activation is increased in the post-condition and another time only brain regions, where activation is decreased in the post-condition. I can’t find any possible programming option in SPM or MATLAB how to separate images showing increased from images showing decreased activation in the SPM glass brain after a paired t–test analysis?
Kind regards and many thanks for helping me!
Dipl.-Ing. Andreas Jell
Heinrich-Heine-Universität
Institut für Experimentelle Biologische Psychologie Gebäude 23.03
Universitätsstraße 1
D-40225 Düsseldorf[fon] +49 (0)211 81-13958
[fax] +49 (0)211 81-14522
[email] [log in to unmask]