Dear SPM users,
I’m currently trying to implement a PPI analysis in my pipeline, but I’m having some trouble in moving from the example/exercise on SPM manual to my data.
I’m working on a fMRI dataset coming from a longitudinal training experiment: each subject performs a given task twice, 9 days apart. Each fMRI session (event related design) has 4 runs. During each run participants are presented with objects and words in pseudorandom order, performing a simple one back task.
The group effect reveals that in both fMRI sessions, the presentation of objects is associated with increased activity in visual and acoustic areas. I have reasons to hypothisize a change in connectivity between these two ROIs on the second fMRI session compared to the first one, thus the need to implement PPI.
I have several basic doubts (first time I work on these procedures):
1. so far I modelled my pre-learning and post-learning data separately, performing repeated measures tests to look for univariate changes. Do I need to put all my 4+4 runs in a single model, for the PPI analysis?
2. in the example on SPM manual the analysis is performed on a single subject and a single run. How do I create the PPI variable when I have multiple runs? I’m referring in particular to ch.33 par 33.5. At point 5 it’s explained how to input variables and contrast weights. In the example, the matrix specified is [2 1 -1; 3 1 1] where 2 and 3 refers to the conditions of interest, while the last column, -1 and 1, are the actual contrast weights. Should I just put, in the first column, the progressive number of where my condition of interest (“objects”) is in the big design matrix, and then just put a contrast weight as -1 and 1 basing on pre-learning or post-learning fMRI session? And I’m assuming to use a big general model, with 4+4 runs (pre learning + post learning);
3. when I try to extract the VOI for each subject, after adjusting for effect of interest, I get the following message error:
Error using uicontrol
While setting the 'Max' property of UIControl:
Value must be numeric or logical.
Error in spm_input (line 1286)
h = uicontrol(Finter,'Style','Pushbutton',...
Error in spm_regions (line 117)
s = spm_input('which session','!+1','n1',s,s);
Error while evaluating UIControl Callback
I don’t get whether this is something related to the number of runs or something else: do I need to extract a VOI for each run in each subject? still, I’m not given the possibility to select the proper run, and if that were the case, I'm even more confused than before on how to treat these VOIs from multiple runs.
I apologise for the long post but I’m kind of puzzled on how to implement the correct procedure. Also, should you have any recommendation for a PPI tutorial covering more complex designs (e.g. multiple runs; group effects; longitudinal studies) feel free to share them!
Thank you,
Simone
|