Christian
> Dear SPM experts,
> I’m a SPM beginner, doing a fMRI study with two groups (19 patients & 19 controls). I am using a block design with three different conditions (condition1, condition2 and rest). Right now, I’m doing the 2nd level analyses and I have met a problem that is quite confusing me.
> First, I have done the within group comparisons (FWE p=0.05). For the comparison condition1 vs. rest the controls show a very high t-value, T=17.44, in one special region (Temporal Pole Sup left; -48 8 -2; cluster size 525 voxels). However, the patients show absolutely no activation in this region. So, I have expected, that if the controls have such a high t-value whereas the patients show no correspondent t-value for the same area, the between group comparison controls vs. patients has to produce a significant result in this area, indicating a much higher activation for the controls. But the between group comparison - FWE corrected as well as uncorrected! - shows absolutely no group differences in this area. This is not what I have expected and I can’t explain this result. I hope that someone of you has an idea…
assuming all is ok with your images, the simplest explanation is that
your patient do have activations but below threshold and the difference
is not significant - a nice way to look at this is to display the con
images corresponding to your T-test (T = con / variance). For instance
use check-reg and select the con images of the one sample t tests of
group 1 and of group 2 - If you see big bright spots at the same
location, then here you go - both groups show activations but patients
don't reach significant level, either because the mean is too low or the
variance is very high.
now it is also possible that something is not quite right with your
images - again an easy way to check this is to look at the masks created
during the stat analysis (mask = area where stats are performed) - using
check reg display the mask images created during the one sample t-test
for group 1 and group 2 - the 2 sample t-test will be performed only in
areas where both masks match (could be that one patient data is not well
normalized and the whole mask is biased removing this part of the brain
from the analysis .. )
>
> A further question refers to the above mentioned t-value. The t-value T=17.44 seems to be extraordinary high to me – especially compared to all the other t-values in my analyses. So, I checked the t-values for every single control subject (n=19) for this special region, in order to get an explanation for this very high t-value. The result was quite confusing, because the single subjects only showed t-values ranging from -0,22945 to 1,05096. This range doesn't explain T=17.44 to me. I hope, that someone of you has an explanation for that high t-value…
for a given subject T = con / variance between scans and this variance
can be quite high
for the group T = con / variance between subjects -- here the con image
is the mean of the con images from all the subjects not the mean from
spmT images .. in addition the variance between subjects can be lower
than the variance between scans
Hope this helps
Cyril
--
Dr Cyril Pernet,
Academic Fellow
Brain Research Imaging Center
http://www.bric.ed.ac.uk/
Division of Clinical Neurosciences
University of Edinburgh
Western General Hospital
Crewe Road
Edinburgh
EH4 2XU
Scotland, UK
[log in to unmask]
tel: +44(0)1315373661
http://www.sbirc.ed.ac.uk/LCL/
http://www.sbirc.ed.ac.uk/cyril
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|