Hi Olivier,
I expect your confusion is due to a change in the way things work
between SPM2 (or older) and SPM5.
Under SPM2, the images are first normalised to the template, then
segmented. So there is some interpolation before segmentation (though
interpolation is not usually that disastrous...).
Under SPM5, the unified segmentation model effectively segments the
images in their native space, and then warps these to template space.
So uninterpolated images are segmented, but there is interpolation
necessary for warping the segmentations. The unified model appears to
warp at the same time as segmentation, but in practice, the priors are
warped to the native image in order to segment it, and then the
segmentations are warped to template space using the inverse of the
transformation used to warp the priors. Hope that's clear.
Best,
Ged.
> My understanding is that the segmentation of GM. WM,… should be done on
> the original structural without any interpolation. The atlas used for
> prior is registered and the whole brain is normalized to the atlas GM
> map, but it is the atlases that are interpolated and the actual EM is
> performed on the original T1W data. Once the deformation matrix and
> fields are computed, they are applied to the results of the segmentation
> (c1, c2,…) to generate probability maps in the correct Tailarach
> coordinate system.
>
> From the code, documentation, archive and papers that I have read, it
> is not clear to me that this is what happens. It seems that the T1W is
> warped/registered and then segmented (or at least the warping is done at
> the same time). If this is the case I would be concerned about
> interpolation artifacts affecting the results of the segmentation.
>
> Could someone clue me about this issue, (if it makes sense at all)?
>
>
> Olivier
>
> __________________________________________________
> Olivier Salvado, PhD - _website_
> <http://ccir.uhrad.com/faculty/oliviersalvado.asp>.
>
> Case Western Reserve University | Case Center for Imaging Research
> University Hospitals of Cleveland | Department of Radiology | Wearn B49
> 11100 Euclid Av. | Cleveland, OH 44106
> Ph. (216) 983 3426 | Fax: (216) 844 4987
>
>
>
>
|