> Should I use my naked eyes to talk about the (de-)activation BA regions
> or use the p-values adjusted for search volume?
I think you are asking about how to determine which Brodmann areas to report for a given cluster of activation. I find that xjview (http://www.alivelearn.net/xjview8/) is very helpful for this. With xjview you can get a report on how many voxels within each cluster are in each BA (and other anatomical labels) using the Talairach Daemon. This is a good alternative to either reporting only the BA for the peak activation or just eyeballing it.
> What should I do with data if the p value of set level is 0.1 (ie) but the
> cluster-level is 0.000? or 0.1 for cluster-level but 0.000 for voxel level?
You need to decide what type of inference you are interested in. A set-level inference only lets you conclude that there is significant activation somewhere in the entire set of voxels you are analyzing. A cluster-level inference lets you conclude that there is significant activation somewhere within the identified cluster. A voxel-level inference lets you conclude that there is a significant activiation in that particular voxel.
|