Russ,

> If there are other reasons to do the transform that overcome these
> objections I hope that someone will point them out.  Given all the
> problems with the original Talairach space it seems that we should
> just embrace the MNI space

I'm sorry to be joining this discussion a bit late. I absolutely
agree that there are many problems in trying to relate the MNI to the
Talairach space. As Matthew pointed out, however, "embracing MNI space"
would mean giving up on the idea of localizing activations in terms of
Brodmann areas.

It seems to me, however, that identifying Brodmann areas is crucial, and
just as important as idenfying anatomical gyri (although this may
be different in different regions of the brain). But for a region
such as the prefrontal cortex, for example, most functional subdivisions
are defined *both* in terms of anatomical gyri and brodmann areas. By
looking at the subject's anatomy, we might be able to tell that an
activation is localized in the middle frontal gyrus, for instance. We
would need to refer to the Talairach atlas, however, in order to determine
if this activation is "ventrolateral" (area 45), "dorsolateral" (area 46),
or "rosterolateral" (area 10) - information that would make A LOT of
difference functionally.

I short, I really believe we need an MNI to Talairach conversion, and the
better mapping we can get, the better off we would be. Matthew and I have
been working on an improved transform, as he's mentioned several
times. What would be really needed eventually is an averaged and
normalized cytoarchitectonic MNI-like template (with information about the
probability of each voxels belonging to certain Brodmann area). It seems
though that it would take a while to arrive at such a template, so until
then, we should probably at least try to make our reports of functional
locations (Brodmann areas and gyri) as accurate as possible.

Best wishes,

Kalina


On Thu, 29 Mar 2001, Russ Poldrack wrote:

> I'm not sure if this is a discussion that's already been had on the
> list, but I worry about using the MNI-Talairach conversion. In
> particular, what's the logic for computing the converted coordinates?
> There are several possible reasons that I could imagine doing this,
> and I think that in each case there are arguments against it.
>
> 1.  to use the Talairach atlas to determine brodmann's areas.  the
> folly of this has been discussed many times on the list.
>
> 2.  to use the talairach atlas to determine the anatomical location of
> an activation.  This also seems to me to be a bad idea, since we
> almost always have the subject's anatomy and can look at it to
> directly figure out where we are (using a high-resolution atlas like
> Duvernoy or Mai et al. to determine what is what if necessary).
>
> 3.  To calculate numbers to put into a foci listing in a paper.  It's
> not clear to me why this is a good idea either.  Although the
> talairach space has been the standard, it seems that just as many
> people now are using the MNI space.  In addition, so far as I know
> Matthew's conversion algorithm has not been fully explored or
> validated (this would be difficult since we don't have imaging on the
> subject from the Talairach atlas).  All the more reason to stick with
> a space that has been well-described and has good templates (the
> MNI305).
>
>
> cheers,
> russ
>
>

_____________________________________________________________________________
Kalina Christoff                      Email:  [log in to unmask]
                                      Office: Rm.455; (650) 725-0797
Department of Psychology              Home:   (408) 245-2579
Jordan Hall, Main Quad                Fax:    (650) 725-5699
Stanford, CA 94305-2130               http://www-psych.stanford.edu/~kalina/
_____________________________________________________________________________