To be honest, I do not know which is likely to be the better option -
particularly as what works best for one dataset may not do so well on
another. The easiest way to find out would be to do the experiments
to see.
If you use the "New Segment" of SPM8 (or just "Segment" in SPM12),
regions of the image that are set to zero will be ignored in the
computations. There is no official lesion masking approach, but this
(I hope) should work.
Best regards,
-John
On 23 May 2013 11:11, Pablo Ripollés Vidal
<[log in to unmask]> wrote:
> Hi SPMrs!
> We have some stroke data that we want to analyse. For the
> normalization, we want to use Unified Segmentation with CFM, as these
> patients have big lesions. We have scanned both a T1-weigthed (1x1x1
> mm3) and a FLAIR (0.5x0.5x1 mm3). The thing is that, in order to
> identify the lesions, depicting them on the FLAIR is much easier.
> My question is, would it be OK to depict the lesions on the FLAIR and
> use this same FLAIR image as an input to Unified Segmentation to
> derive the normalization parameters?
> Or would it be better to create the lesion masks on the FLAIR,
> corregister and reslice these masks onto the T1 weighted space and
> then use the T1 with Unified Segmentation?I am thinking on this, as if
> I am not mistaken, SPM8 has the ICBM 452 as a template, which is an
> average of T1s and not FLAIRS.
> Thanks a lot!!
>
> --
> Pablo Ripollés Vidal
> Ph.D Student
> Cognition and Brain Plasticity Unit (Universitat de Barcelona)
> http://www.brainvitge.org/
> IDIBELL [Bellvitge Biomedical Research Institute]
> Currently at the Otto-von-Guericke University
> Abteilung für Biologische Psychologie
> http://www.ipsy.ovgu.de/Abteilungen/Biologische+Psychologie.html
|