Hmm I looked at the html for BET, the template_4D_GM and the GM_mod
merg timeseries (both in fslview) and they all look fine. The results
are outside of template-4D_GM and GM_mod_merg highlighted regions as
well as the MNI152 brain. The MNI152 brain is very closely aligned
with the template_4D_GM and the GM_mod_merg files. The protocol I used
was as follows:
fslvbm_1_bet -N; fslvbm_2_template –n; fslvbm_3_proc –T; cd stats;
randomise -i GM_mod_merg_s3 -o Discont_Regress_HAMD -d HAMD.mat -t
HAMD.con -e HAMD.grp -T
So I think that I am using the correct protocol.
Thank you Gwenaëlle for the quick response,
Jeff
On Tue, Aug 17, 2010 at 12:28 PM, Gwenaëlle DOUAUD
<[log in to unmask]> wrote:
> Hi Jeff,
>
>> I have been running FSL VBM analyses and many times I have
>> my most
>> significant results appear to fall outside of the actual
>> brain
>> (somewhere the skull should be according to an overlapped
>> standard
>> MNI152 brain). Why would this be? Is it possible that
>> fslvbm_3_proc is
>> not scaling my nifti files appropriately? or is it finding
>> a
>> difference in skull density? or maybe something else is
>> happening?
>
> At this stage of the analysis, there should not be any skull in your images. If there is, it means that you have not followed the fslvbm protocol or you have not checked properly your images after the fslvbm_1_bet.
> A good thing might be to look at the GM_mod_merg 4D image in fslview and check each volume to see if the registration is correct, which I suspect is not... If this is indeed the registration that has failed, you will need to determine why by checking each step of the protocol up until the registration.
>
> Cheers,
> Gwenaelle
>
> --------------------------------------------------------------------
>
> Gwenaëlle Douaud, PhD
>
> FMRIB Centre, University of Oxford
> John Radcliffe Hospital, Headington OX3 9DU Oxford UK
>
> Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717
>
> www.fmrib.ox.ac.uk/~douaud
>
> --------------------------------------------------------------------
>
>
>
>
|