Hi,
New-segment is specifically sensitive to misalignment of images with TPMs. So as the first step I would check the registration between the problematic image with TPMs. I would also check the header to see if it's the same as other images that led to correct segmentations.
All the best,
Arman
Sent using a mobile device, please excuse my brevity/misspellings.
On 13 Jun 2012, at 10:29 PM, "N.E. van der Aa" <[log in to unmask]> wrote:
> Dear SPM users,
>
> I'm currently experiencing some difficulties segmenting a T1 weighted image. The particular MRI was acquired of a 12-year old child, and I therefore created a pediatric template and TPM using Template-o-matic. Using this template, I have been able to segment a large number of MRIs using the VBM8' segmentation function ('Estimate and Write'), resulting in normal segmentations.
>
> For this particular child however, something seems to go wrong. Attached you can find four figures. Figure 1 shows the dartel warped, modulated T1 (wmrAnatom.nii). Figure 2 shows the original Anatom.nii after applying the deformation field (y_rAnatom.nii). As you can see, some parts of the cortex are missing in the wmrAnatom (fig1). This becomes more obvious when the cortex segmentation (wrp1Anatom.nii) is overlayed (fig 3). Obviously, when overlaying the segmentation in native space, the same can be observed (fig 4).
>
> So my questions is, what may cause this problem? I have tried the New Segment, which resulted in a similar segmentation, while the 'standard' Unified Segmentation gives a normal results (i.e. including the entire cortex). I have played around with some parameters (warping regularisation, sampling distance), but with moderate results. I have checked the origin of the image, which is normal and there seem to be no artefacts which might explain any problems. Any suggestions which may improve the results would be more than welcome!
> Thanks,
>
> Niek van der Aa
> <fig1.jpg>
> <fig2.jpg>
> <fig3.jpg>
> <fig4.jpg>
