Dear Dr. Jenkinson,
Thank you very much for your input.
I think I have a greater issue with the brain extraction process than the
registration itself at this stage. When I launch BET, I put FA=0.25 to make
sure that I get the most brain possible. However, BET doesn't seem to
consider the extensive lesion as part of the brain, so there is a big chunk
of the brain missing (like a shark bite) corresponding to the area of the
lesion. I was told to put a mask on my lesion, but I am afraid that I need
some guidance regarding the next step.
Again, thank you very much for your help,
Jean-Sebastien
Le 4/3/12 8:20 AM, « Mark Jenkinson » <[log in to unmask]> a écrit :
> Dear Jean-Sebastien,
>
> It is difficult to know might be going wrong without knowing more.
> A complete failure, with very wrong orientation, is not uncommon when things
> go wrong. However, something with the right "orientation" but poor
> registration
> is really no better and you should not use either.
>
> It is probably best to use the brain extracted version if you have this and
> the
> brain extraction is good. In that case I would try the following options with
> FLIRT
> (in various combinations):
> -usesqform
> -nosearch (you may already by using this)
> -cost mutualinfo
>
> If none of these (or combinations of these) help then let us know.
>
> All the best,
> Mark
>
>
>
>
> On 2 Apr 2012, at 17:38, Jean-Sebastien Provost wrote:
>
>> Dear FSL experts,
>>
>> Like many, I'm trying to register into the standard space scans with
>> extensive lesions. So far, I have followed the instructions that was
>> previous given in another email, and they have been working well. Basically,
>> I have created a mask by manually drawing a ROI over the stroke area,
>> inverted the mask with the following command "fslmaths originalmask -mul -1
>> -add 1 -thr 0.5 -bin invertedmask", and registered using FLIRT and putting my
>> inverted mask into the weighting option as an input.
>>
>> Like I said, it has been working well for 2 subjects, but didn't for 2
>> others. In the latter case, when I look at the data in FSLview, the image is
>> no longer aligned: it has suffered a severe rotation making the axial view
>> now the coronal view with a 90 degree rotation, and so on for the other
>> views.
>> Now, I also tried to do the registration with a betted anatomical scan. Here
>> the orientation is fine, but the registration itself doesn't look good.
>>
>> Any suggestions would be useful.
>>
>> Thank you in advance,
>> Kindest regards,
>>
>> Jean-Sebastien
>>
|