Dear Samantha,
Just in case, as you were refering to some scanner problems / signal drop, I would suggest to have a closer look at data quality, e.g. there might be scanner spikes or other artefacts well before the more salient ones. The movie tool in Artrepair is very useful for that purpose. If there's really some issue with something like overheating it might also well depend on how much the scanner was used before your subjects. E.g. the morning subjects might be alright, but those in the afternoon might show a signal drop much earlier.
With regard to significances on single-subject level, to add on what Colin has already stated: It can be useful (and in fact more useful than looking at the thresholded T maps) to load e.g. all the con_0001 images (which might represent e.g. 1-back vs. 0-back) with "Check reg". Right-click on one of the images, go with Image \ Window \ global \ manual, then enter something like -5 5. Which exact range to use depends on the data, the main idea is to threshold the different con images the same way. You could now check whether the beta images look reasonable (e.g. bilateral increases in working memory networks), whether there are any weird patterns (might result from e.g. fast head motion). For a more objective approach you could also z-transform the con images & search for outliers.
Best
Helmut
|