Dear Amandine,
In the fslmaths command you are taking the mean over *all* voxels, not
just the non-zero ones in the skeleton. So this is why you get such small
values - it is including all the zero voxels outside of the skeleton too.
The fslmeants uses a mask and so gives you the correct result.
For your interest, if you ran your mask through the same fslmaths command
and then divided your number by the number you obtain using the mask only,
then this would renormalise the result and give you the correct mean within the
mask.
All the best,
Mark
On 2 Sep 2011, at 15:03, Amandine Pelletier wrote:
> Dear FSL experts,
>
> I'm working on a cohort of 154 subjects. I would like to extract mean FA value on the skeleton for each subject. So, I performed the command line below:
> fslmaths all_FA_skeletonised.nii.gz -Xmean -Ymean -Zmean global_FA_skeletonised
>
> Then, in order to extract FA values on the skeleton, I used the command line below:
> fslmeants -i global_FA_skeletonised -o meants.txt
>
> I think that FA values on the skeleton are crushed by the FA values out of the skeleton, because I obtain mean FA around 0.0075-0.0080 while I used an other method using the command line below:
> fslmeants -i all_FA_skeletonised.nii.gz -m mean_FA_skeleton_mask -o meants.txt
> I obtain mean FA around 0.46.
>
> Why such a difference ? Why I don't try approximately values around 0,45 with the first method (fslmaths -X mean -Ymean -Zmean).
>
> Thanks for your time and help,
>
> Amandine
>
|