Hello Angela and Alessandro,
I was just working on the approach of Steffener et al. (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2818488/) which is based on the Calhoun paper and found it quite useful.
The way I understand it, you actually have to build t-contrasts at the first level. It's not explicitly stated, but I figured it from the contrast vectors/weights in Table 1 of the publication. Usually F-contrasts need to be in matrix form like [1 0 ; 0 1]. If I'm wrong, then somebody please correct me. As Alessandro stated, you can then calculate the magnitude for each subject and take this magnitude image at the second level and do a group analysis using a t-contrast. By using the contrast weights you can then create masks that limit the activation to BOLD responses before that canonical time-to-peak of 5 s and after it; basically you can look at responses from 4-5 s or 5-6 s or 4-6 s, depending on the mask.
I found the scripts provided by Jason quite useful. You can find them here: https://sites.google.com/site/steffener/Papers
Regards,
Glad
|