> i went to the link john mentioned
>
> > http://www.bic.mni.mcgill.ca/cgi/icbm_view/
>
> when i entered Y=2 and Z=-3 or -4 and updated the image then i think one
> finds a much better localization of the ac (just try to click next to the
> brain so that the cross hair does not block your view). or at least i think
> that's where the ac should be. could sombebody please comment if this is
> correct? and even if it is whether this 3.5-4.5mm transposition of the
> origin makes any difference in general for normalization etc.? thanks in
> advance,
My understanding is that the ICBM space was devised by 9-parameter registration
of manually defined landmarks in subjects brains, with the Talairach & Tournoux
Atlas. The space is the closest affine transform such that the points are
matched. As the T&T atlas was based on a rather distorted brain, then moving
the space so that the AC was in the right place, would mean that the other
landmarks would not fit together well. A compromise was needed so that the
landmarks were as close as possible to where they should have been (probably
by minimising the sum of squared distances of all the distances).
Very few discrete points can be found within a brain, and different choices of
landmarks give different solutions - especially when registering to a distorted
standard. More sophisticated registration techniques may have given better
results (averages of landmark registered images tend to be very blurred compared
to images registered by minimising some cost function such as mean-squared
difference), but the original space was devised some years ago.
I originally believed that the ICBM space would become a standard. However,
more recently, at least two more ICBM spaces have arisen:
http://www.loni.ucla.edu/ICBM/ICBM_452T1.html
http://www.loni.ucla.edu/ICBM/ICBM_Probabilistic.html
Best regards,
-John
--
Dr John Ashburner.
Functional Imaging Lab., 12 Queen Square, London WC1N 3BG, UK.
tel: +44 (0)20 78337491 or +44 (0)20 78373611 x4381
fax: +44 (0)20 78131420 http://www.fil.ion.ucl.ac.uk/~john
|