Hi,
We don't generally recommend using priors for segmentation,
particularly for cortical regions, as the priors are too blurred to
be useful and the dangers of misalignment are messy.
However, for subcortical segmentation, priors can help where the
contrast is weak, and the integrated registration and segmentation of
SPM5 probably works well on the thalamus, so this may give you the
best thalamus segmentation at present.
The next release of FSL will contain a new tool for subcortical
segmentation which uses a completely different method, namely mesh-
based modelling of the different sub-structures, with shape and
intensity information learned from rich training datasets.
Cheers, Steve.
On 11 May 2007, at 16:59, Wayne Su wrote:
> Hello,
>
> These days I have been confusing about the thalamus segmentation.
> Thalamus
> is a deep gray structure. I used FAST with -A parameter on 1.5T T1-
> weighted
> scans. The result is quite good for most of them. But for the 3T T1-
> weighted
> scans, the result is not satisfied. I tried the segmentation from
> SPM5 on 3T
> scans, the result is good. So I am quite interested on how to
> improve the
> result from FAST. I also tried -b and --b parameters but the
> improvement is
> limited. Could you please give me any idea what's different between
> FAST and
> SPM's. Thanks.
>
> Wayne
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|