Dear SPMers,
Sorry to be asking for your help again regarding this issue but I have been unable to decipher it yet.
As an example, the context being that I am studying the contrast : negative vs neutral, do a second level analysis with the con images related to that specific contrast and run a one sample t-test.
Then, I use marsbar to know the values for a specific ROI.
I have approached that in 3 ways:
a)set the design from the second level SPM.mat file and get the values from the con images for that specific ROI.
b)extract the percent signal change for all subjects from the first level SPM.mat and get the ROI values for the conditions (including the negative and the neutral conditions) (averaged values from the smooth files). With that info I:
b.1) Do a substraction between conditions
b.2)Regress the neutral condition from the negative and saved the adjusted predicted values.
If I do correlations between this 3 ways of getting information for a specific contrast for an ROI, they are not significant.
To note, if I correlate the second level t-test “negative vs baseline” with the % signal change of the negative condition, the correlations are high.
So I think that the issue relies on how I compute the contrast information after I have extracted the percent signal change from each of the conditions, if I don’t want to gather the information from a second level SPM.mat.
I also want to note that I have ordered the conditions the way marsbar orders them, so I am almost sure that I have the correct information for the conditions.
Thanks tons everyone in advance,
Best,
V
|