I am working on a event fmri, emotional memory paradigm. I have 7 subjects,
each having 8 different events (positive hits, negative hits, neutral hits,
positive , negative and neutral misses, correct rejections and false alarms).
I have generated 8 contrast images using first level design for each
subject. (condition vs. baseline). Now I want to generate second level analysis.
My intention is to create positive hits minus negative hits,
negative_positive hits, collapsed emotional hits - neutral hits, all correct
hits - correct rejections. (for the latter 2 contrasts I want to use
inclusive masking with the former 2 contrasts.)
I have two options for second level analysis:
1. what I call as 'multiple SPM.mat' option:
create first level contrasts for collapsed and substraction outcomes, do
individual one sample t test by putting together all 7 subjects at latter
level. - is this a valid model?
The issue here is SPM5 does not allow me to use masks generated from one
design in another. Can I work around this using ImCalc options?
2. Flexible factorial option: Here I create a single spm.mat with the
following design.
factors: 1. subject (equal variance) 2. conditions (unequal)
enter all 8 contrast scans/ subject by creating 7 subject entries. the
condition in each subject entry is similiar [1 2 3 4 5 6 7 8]
main effects 2X1 model
I get this matrix (see attached)
Why is the condition parameter uniformly white? I expected boxcars in it for
8 different conditions. Is it valid to proceed with this design matrix to
generate contrasts such as (0 1 1 -2 ) - for collapse emotional vs neutral hits?
Any help is very much appreciated.
|