Thank you so much for taking the time to look into this and for your
advice Jesper - it is very appreciated!
Best,
Stephane
Stéphane Jacobs - Chercheur post-doctorant / Post-doctoral researcher
ImpAct - Inserm U1028 - Equipe Pélisson
Centre de Recherche en Neurosciences de Lyon
16 avenue du Doyen Lépine
69676 Bron Cedex, France
Téléphone / Phone: (+33) (0)4-72-91-34-20
Le 10.07.13 13:32, Jesper Andersson a écrit :
> Hi again Stephane,
>
> I have looked at your data and the problem is, as we see increasingly
> often, high intensities in what we believe is marrow in the skull. For
> some reason this is something we mostly see in the top slices (i.e.
> the crown of the head) so this is where the warps typically go wrong.
>
> It is important to realise that for non-linear registration we rely on
> the images (the template and the registree) to be reasonably similar,
> and that this is true also for extracerebral tissue when it is located
> closely adjacent to the brain. If you look at the top slices of the T1
> template you will set that as you go outwards from the brain the
> intensities drop, i.e. brain->high_intensity and
> outside_brain->low_intensity. If you look at your scan you will see
> the opposite. This is something that is very hard for an intensity
> based non-linear registration to deal with and you will get the kind
> of problems that you see.
>
> We are aware of the problem, and will hopefully come up with a
> solution at some stage.
>
> BUT, for the time being the solution is to make sure that the
> structural scans you acquire resemble the T1 template. If your
> structural scans don't, I suggest you chat to your local MR physicist
> to get some advice on what sequence to use. It is typically a good
> idea to use some level of fat supression.
>
>
> Stephane, in your specific case it turns out that bet does a good job
> and if you run
>
> /usr/local/fsl/bin/fnirt --in=highres --aff=highres2standard.mat
> --cout=highres2standard_warp --iout=highres2standard
> --jout=highres2highres_jac --config=T1_2_MNI152_2mm --ref=standard
> --refmask=standard_mask --warpres=10,10,10 --imprefm=0 --impinm=0
>
> you will get good results. What you are doing here is to use instead
> the betted scan AND turn off implicit masking.
>
> Jesper
>
>
>>>>>> Hello,
>>>>>>
>>>>>> I use FNIRT with default settings to register my anatomicals to the standard space, and am not completely happy with the result for some subjects.
>>>>>> The calibration of functional data to the highres anatomical is done using the BBR option, and woks fine. But when I look at the highres2standard output, I see some local distorions, especially in the parietal region, i.e. some spikes poking outside of the standard template. These distorsions are of course passed on to the example_func2standard registration, and the output shows the same deformations. I have put the images for the 3 registrations here:https://drive.google.com/folderview?id=0B6BVou-GmdvuNmNIdnd3eGs1MW8&usp=sharing.
>>>>>>
>>>>>> Could someone have a look and tell me what I can do to fix this?
>>>>>>
>>>>>> Many thanks for the help, and best regards
>>>>>>
>>>>>> Stephane
>>>>>>
>>>>
>>>
>>
>
|