Thank you Stam! On Tue, May 31, 2016 at 5:22 PM, Stamatios Sotiropoulos < [log in to unmask]> wrote: > Hi Mahmoud > > The topic of FA values larger than 1 or negative eigenvalues of the tensor > and the fact that it does not influence group-wise comparisons have been > discussed many times in this list. In short it just comes from the fact > that the tensor is estimated using linear least squares, which do not > impose any constraints on tensor eigenvalues positivity. > > Cheers > Stam > > > > On 31 May 2016, at 14:17, Mahmoud <[log in to unmask]> wrote: > > > > Hi All, > > > > I've noticed that there are values in FA maps (mostly after convolution > with a kernel) that are higher than 1 and also very small values in the > background ( < 0.01) which are obviously not correct. > > I was wondering if it's a safe practice to threshold my FA maps like > below before further processing: > > > > fslmaths FA_map.nii.gz -thr 0.01 -uthr 1 FA_map.nii.gz > > > > Thank you! > > Mahmoud > > >