Thank you Mathew, I didn't attempt a final final thresholding/binarising step . Rather I switched to fslcreatehd and found I could create the binarized sphere mask but I will try this with the next mask. Thanks again On Mon, Sep 22, 2014 at 11:42 AM, Matthew Webster < [log in to unmask]> wrote: > Hi Charlie, > You have created a single voxel with intensity one with > the first fslmaths call, which is then filtered ( smoothed ) across a > 15-voxel radius sphere, resulting in values less than 1 in the sphere. The > values outside the sphere are numerical noise. When creating masks, it can > often be worth applying a final thresholding/binarising step to set the > intensity of all masked voxels to 1. > > Kind Regards > Matthew > > FSL group, > > > > I have made a sphere mask centred on significant activation coordinates > (from a previous TBSS analysis). > > fslmaths $FSLDIR/data/standard/MNI152_T1_1mm -mul 0 -add 1 -roi 116 1 > 101 1 88 1 0 1 cst_slfroi -odt float > > fslmaths cst_slfroi -kernel sphere 15 -fmean cst_slf_sphere_mask -odt > float > > > > My question concerns the intensity values with the sphere. If the -add 1 > command adds the value 1 to the region within the sphere, then why is the > value in the sphere 7.1e-05, a value far less than 1. Also, the value > outside the sphere is not zero. In short, this mask must require additional > thresholding and binarising. > > > > While I can combine atlas based masks to cover the area of interest, the > combined masked is too dispersed. Any suggestions with the sphere mask > would be much appreciated > > > > All the best, > > charlie >