Hi Sarika,
you seem to touch upon several issues here. Let's see if I can comment
on some:
> I would like to know which of the following methods serve better
> for normalization of imaging data from children.
> (1) Template created with TOM toolbox with the NIH dataset, by
> specifying the details of the participants (age, gender etc)
> (2) Using the individual structural T1 to create sn.mat through
> segmentation and later use it for normalizing.
In order to be able to do this, you will first need to define what is
"better normalization", a question I always found difficult to answer.
You should also make sure you have similar power for either approach,
which I do not believe (see below).
> I would like to compare results from different age groups and hence i am
> concerned about using different templates for each age group.
Concerning participants from different groups is always tricky. If you
compare younger children (say, 6-8) with older children (say, 12-14), an
alternative may be to use a template that is "in the middle" so that
both groups will be treated similarly. This is what TOM allows you to
do. If the difference between them is too big, the problem may become
really big.
> Is there a way to create a template using the structural data from my
> own dataset?or is it better to use method (2)
There are, of course, ways to do this, but in your case, with n=15, I
would definitely go with option 2. If you check the figures in the TOM
paper, you can clearly see that the quality of the resulting priors is
clearly dependent on the number of contributing subjects. We chose n =
12, and the results are not pretty. For such small groups in particular,
TOM was created.
> Also, how do i interpret the co-ordinates since it is pediatric brain?
> From the mailing list I understand that mricro might not work well for
> children's data. Is there an alternative??
This is not a problem of MRIcro(n), but of the fact that the pediatric
brain is different from the adult one. Even if you believe in strict
Brodman's areas (which I don't) in adults, these cannot be expected to
be fully-formed in children. The coordinates would therefore not give
you very meaningful information even if you used an adult template for
spatial normalization. If you use a pediatric template, this issue is
aggravated even further.
> The participants in my study are in the age group 8-9 (n=15) and I use
> SPM5 for analysis.
Well, it seems the age range is narrow enough, but again, the groups
size, in my humble opinion, is too small to create your own meaningful
template. It is up to you to compare the two approaches, but keep in
mind that priors based on your own subjects will always be more
appropriate as they are, well, your own subjects. In order to properly
compare these approaches, you would also need to have 15 matched kids
that are not part of your study; you could conceivable use some of the
NIH data for that.
Hope this helps,
Marko
--
=====================================================================
Marko Wilke (Dr.med./M.D.)
[log in to unmask]
Universitäts-Kinderklinik University Children's Hospital
Abt. III (Neuropädiatrie) Dept. III (Pediatric neurology)
Hoppe-Seyler-Str. 1, D - 72076 Tübingen
Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473
=====================================================================
|