> I'm confused about whether it is correct to do the ROI analysis on these
> activation maps or on the pre-fimmed functional datasets.
Different people mean ROI analysis in different things. If you simply want to plot out the ROI data, you can extract the data based on whatever way you prefer. However, if you want to run real statistical analysis on an ROI, some people argue that it's better to go with an independent definition of ROI from your activation map to avoid circularity issue, e.g., anatomical ROI or region from an independent study.
> Does it make sense to generate ROIs from a built-in atlas and then use
> 3dmaskave to perform the analysis on the tlrc'ed fims, or do I need to go back
> to the original files?
Built-in atlas seems fine to me.
Gang