 |
 |
Sam, There isn't a contrast that can be generated to capture #1 or #2, so there isn't a map that could be brought to the group level. Generally, conjunctions are computed at the group level. An alternative, in your case, could be to count the number of subjects that have the conjunction at the 1st-level. Best, Donald Best Regards, Donald McLaren, PhD On Thu, Mar 3, 2016 at 5:41 PM, Sam Torrisi <[log in to unmask]> wrote: > hi folks stronger in stats than i- > > We want to make a "connectivity conjunction map" in the sense that we have > resting state correlation maps with two different seed regions and we want > to see: > > (1) what is uniquely connected with either region > (2) what is mutually connected with both regions > > i think we've figured out #1, but for #2... it can feel incredibly > arbitrary to just threshold each group map individually and then calculate > an "AND" conjunction. firstly, does the correction stay the same at the > higher level (i.e. if each map was separately thresholded at p<0.05 > corrected is the conjunction also p<0.05 corrected or something else?). > second, the resulting #2 conjunction map is binary, so there's a loss of > info about peaks and variations in mutual connectivity within the clusters > it yields. > > what we'd like is to somehow retain single subject variance up to one > final group statistical map and then threshold only once at the end. that > would yield a normal and more easily reportable statistical map of regions > (with peaks) that are mutually, functionally connected to both regions. > > but is this possible? is what we want not actually a conjunction map? is a > different statistical approach recommended here? thanks so much! > > -Salvatore (Sam) Torrisi > > Section on Neurobiology of Fear and Anxiety > National Institute of Mental Health >