Hi again Eddy,
as I pointed out before, there will still be the problem of substantial distortion between your FA and your T1w...
Jesper is going to reply in much more detail to your specific question!
Cheers,
Gwen
--- En date de : Jeu 30.4.09, Eddy D.L. <[log in to unmask]> a écrit :
> De: Eddy D.L. <[log in to unmask]>
> Objet: Re: [FSL] Re : [FSL] Question about FMRIB58_FA standard-space model
> À: [log in to unmask]
> Date: Jeudi 30 Avril 2009, 13h13
> Hi Gwenaëlle,
> I'll be better make myself clearer on some points.
> The overall TBSS process works fine on FA maps and all
> trasnformation it
> computes from them (basing on the FMRIB58_FA) are then
> applied to the other
> non-FA data.
> Since this process takes into account the most relevant
> parts of the White
> matter,
> I guess one might not be sure that all the transformation
> (either linear or
> nonlinear) that occur, take into account (at least some of)
> the grey matter
> features.
> So my idea was basically to previously perform a grey
> matter alignment with
> the VBM, and to use the warp fields computed as
> "initial guess" for the
> further TBSS process's alignment procedure.
> This initial VBM aligment could be performed just basing on
> T1w images of
> every single subjects (given that they are provided, and
> they are already
> aligned with their respective DwI dataset), or simply
> computing a "fake T1w
> image" by taking the negative of the T2w image of the
> baseline
> Am I completely wrong?
>
> Thanks for your time, by the way!
>
> Eddy
>
>
>
>
>
> >Hi Eddy,
> >I'm not sure I follow 100% of what you're
> saying...
> >The FMRIB58 was created by *linearly* placing in the
> MNI space all the 58
> FA images using >the MNI 152 T1w template. Then only the
> *mean FA* was
> iteratively used to non-linearly >register all the 58
> images.
> >I don't think you should use a GM template in the
> way you describe as there
> are distortions in >the DW images that make some
> structures of your GM
> segmented image and the >corresponding best FA target
> non comparable. This
> is why all the strategies in TBSS 2nd step >(-T or -S)
> are using a
> registration from an FA image to another FA image (either
> the best FA
> >target or the FMRIB58).
> >
> >Hope this helps,
> >Gwenaëlle
|