If you provide the probability volumes to find_the_biggest, it will output what Stam suggests. Peace, Matt. -----Original Message----- From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Stamatios Sotiropoulos Sent: Thursday, March 24, 2011 12:04 PM To: [log in to unmask] Subject: Re: [FSL] ProbtrackX Harvard Target Hi, the atlas is probabilistic, you can access the probability info for each structure in the fsl directory under data/atlases. A straightforward way would be to use these probabilities to set these boundaries, i.e. include a voxel within the region for which its probability is highest. Cheers, Stam On 24 Mar 2011, at 16:39, R. Duke wrote: > Hello FSL Experts, > > I am currently running BedpostX & ProbtrackX and was wondering about using the Harvard-Oxford cortical atlas for my classification targets. My concern is that many of the ROI's overlap (i.e., Superior Frontal/Middle Frontal/Frontal Pole). Would it be appropriate for me to threshold these structures in order to make smaller (non overlapping) ROI's? If so, can someone recommend an objective way of doing this? If I just look at the maps it seems as if a min of 30 minimizes the overlap, however this is not a very objective way of isolating these areas. Any feedback would be absolutely fabulous. > > Cheers > > R- >