In most cases, I'm not entirely convinced that this is worthwhile. In
older versions of SPM, "customised templates" were often used in order
to deal with the wide variety of different types of images that are
collected in different labs. What one group calls a T1-weighted image
is often very different to what another group says is T1-weighted.
These "custom templates" were useful because at the time, the spatial
normalisation part of the preprocessing was done by minimising the sum
of squares difference between the image and template. For this to
work, the contrast in the template image needs to be identical to that
of the images you want to normalise to it.
More recently, the objective function for the alignment has changed to
one of matching tissues. Contrast in the original MR scans is
therefore much less of an issue, although if you are dealing with
unusual populations, there may be some slight bias arising from the
unrepresentative nature of the SPM-supplied tissue probability map
There are a few papers in the literature that describe the generation
of tissue probability maps for unusual populations, which could be
helpful. One such example is:
Population-averaged standard template brain atlas for the common
marmoset (Callithrix jacchus)
NeuroImage, Volume 54, Issue 4, 14 February 2011, Pages 2741-2749
K. Hikishima, M.M. Quallo, Y. Komaki, M. Yamada, K. Kawai, S.
Momoshima, H.J. Okano, E. Sasaki, N. Tamaoki, R.N. Lemon, A. Iriki, H.
There are numerous other examples also out there, some of which relate
to age-specific templates, neonate templates etc. Some of the
examples in NeuroImage include:
Multi-contrast human neonatal brain atlas: Application to normal
neonate development analysis
NeuroImage, Volume 56, Issue 1, 1 May 2011, Pages 8-20
Kenichi Oishi, Susumu Mori, Pamela K. Donohue, Thomas Ernst, Lynn
Anderson, Steven Buchthal, Andreia Faria, Hangyi Jiang, Xin Li,
Michael I. Miller, Peter C.M. van Zijl, Linda Chang
Unbiased average age-appropriate atlases for pediatric studies
NeuroImage, Volume 54, Issue 1, 1 January 2011, Pages 313-327
Vladimir Fonov, Alan C. Evans, Kelly Botteron, C. Robert Almli, Robert
C. McKinstry, D. Louis Collins and the Brain Development Cooperative
The Template-O-Matic: A toolbox for creating customized pediatric templates
NeuroImage, Volume 41, Issue 3, 1 July 2008, Pages 903-913
Marko Wilke, Scott K. Holland, Mekibib Altaye, Christian Gaser
Specific requirements for the tissue probability maps in SPM are that
there should be sufficient types of tissue in order to have a good
model of the data, and that the logarithm of the tissue probabilities
should be finite and spatially continuous. The crispness of the
tissue probability maps should also reflect the accuracy with which
the registration is done within the segmentation.
2011/3/27 Linling Li <[log in to unmask]>:
> Dear all,
> I want to konw how to get the tissue probability maps which are needed for
> segment based on the structural images of a specific group of subjects.
> Thank you all in advance!
> Linling Li
> Research Centre for Neural Engineering
> Phone: +86-755-86392232
> Email: [log in to unmask]
> Shenzhen Institutes of Advanced Technology,
> Chinese Academy of Sciences.