Hi Dylan,
this is not a problem related to the version you're using. This is actually a priori *not* a problem at all.
This happens if you have chosen to use the *FMRIB58 skeleton* for your TBSS analysis (wth the -T option at the tbss_3 stage). This FMRIB58 template and the corresponding skeleton have been created with nreg, the nl registration we were using before in TBSS, and nreg requires the images to be multiplied by a factor 10000.
So yes, a threshold of 2000 seems reasonable and the thresholded skeleton should look fine on your all_FA and not extend beyond them. If it does, then mask it with the mean_FA_mask...
Cheers,
Gwenaëlle
--- En date de : Ven 13.3.09, Dylan Alegria <[log in to unmask]> a écrit :
> De: Dylan Alegria <[log in to unmask]>
> Objet: Re: [FSL] mean_FA_skeleton values around 2000 til 8000
> À: [log in to unmask]
> Date: Vendredi 13 Mars 2009, 0h55
> Hi,
> I am also experiencing the same problem, and the only
> version of fsl on my
> computer is 4.1.2.
>
> I have worked around the problem by setting my threshold
> for the at 2000
> (however, I am not 100% confident that this is appropriate
> because the
> FA_all volumes do have values that range from 0-1).
>
> I'm having difficulty with the previous step as well.
> When I visualize the
> FA_all and FA_mean_skeleton together the skeleton extends
> beyond the range
> of the FA_all stacks. It appears to be delineating white
> matter that would
> be there if the stacks hadn't been masked. Should this
> be happening? And
> will it effect my results? I've attached a photo to
> better demonstrate
> what's happening.
>
> Thanks experts!
>
> Dylan
>
> On Fri, 6 Mar 2009 23:25:56 +0100, Kathrine Skak Madsen
> <[log in to unmask]>
> wrote:
>
> >Well, I guess you could check your path with the
> "which" command, whether
> >it points you to the tbss scripts in fsl-4.1.2/bin/
> >
> >e.g. which tbss_1_preproc
> >
> >Kathrine
> >
> >
> >
> >> Hi Kathrine, I just checked I use fsl 4.1.2 and
> ran the analysis
> >> yesterday. How can I check the TBSS version?
> >>
> >> Michael
> >>
> >>
> >> On 6-Mar-09, at 11:33 AM, Kathrine Skak Madsen
> wrote:
> >>
> >>> Hi Michael,
> >>>
> >>> It seems like your data have been analysed
> with the old TBSS version.
> >>>
> >>> In the old TBSS version (TBSS 1.1, FSL 4.0),
> the FA data were scaled
> >>> to be
> >>> between 0-10000 in tbss_1_preproc, whereas
> there is no scaling (FA is
> >>> between 0-1) in the new TBSS version (1.2, FSL
> 4.1).
> >>>
> >>> Best wishes
> >>> Kathrine
> >>>
> >>>
> >>>> Hi fsl experts, I was just following the
> tbss instructions and wanted
> >>>> to check after <tbss_3_postreg -T>
> if the threshold of 0.2 is good.
> >>>> <fslview all_FA -b 0,0.8
> mean_FA_skeleton -b 0.2,0.8 -l Green>
> >>>>
> >>>> I played around with the threshold in
> mean_FA_skeleton and nothing
> >>>> really changend and then I realized that
> the values range in the
> >>>> thousands.
> >>>>
> >>>> fslstats mean_FA_skeleton.nii.gz -R
> >>>> 0.000000 8676.000000
> >>>> fslstats mean_FA_skeleton.nii.gz -r
> >>>> 2038.859985 7617.527832
> >>>>
> >>>> Any ideas?
> >>>>
> >>>> Michael
> >>>>
> >>
|