Personally I would opt for the atlas approach if possible.
Spherical ROIs based on coordinates do not normally follow the typical anatomy that well.
All the best,
> On 14 Aug 2017, at 10:59, Navot Naor <[log in to unmask]> wrote:
> Hi all,
> I want to create a few masks for an ROI analyses, and i am trying to figure out which way yo go, should i create the mask based on an locating structures in an Atlas (Not all the the regions i am looking for are easy to find in atlases such as the dlPFC), or should i opt for coordinate driven approach?
> If i go with the second option, what size sphere should i use? is there a rule of thumb? or a method of transforming cluster size to sphere size? the largest cluster is 1570 and the smallest is 69.
> Thank you!