Hi Morgan,
See comments below.
Also, have you met Johannes Klein, who also works in the FMRIB? He has
extensive experience with Caret registration.
Donna
On 09/22/2005 06:49 AM, Morgan Hough wrote:
Donna,
Thanks for your email. I was just following the instructions at the top
of this page:
http://brainmap.wustl.edu/help/mapper.html
It's on my do-list to update this page (along with several others), but
was there a page that linked to here? If so, I think I'll remove that
link, because this page is so out-of-date (although still helpful for
those that are still using colin and/or older versions of Caret).
which solicits nominations. The email is set to [EMAIL PROTECTED] which
I guess gets forwarded to the list.
Yeah, John got tired of the spam, so we aliased that address to the
list, which allows only member posts.
That said, I am very interested in using caret to visualize results and
look at the correspondence to probabilistic cyto maps, etc. These data
are 99 FMRI sessions across three tasks collected from a single subject.
It is rather complicated but I am only just getting into caret and your
email is very helpful. So if I am doing all my analysis with FSL tools
and I have a high-resolution structural image for this single subject I
should still FLIRT to MNI 152?
There are some good reasons to spatially normalize before segmenting,
but they typically apply to inter-subject analyses.
One reason that applies to single subject analysis is that it eases
drawing of borders by ensuring AC-PC alignment. Just making sure the
subject is well aligned to AC-PC makes it easier to draw borders during
flattening, because the template frontal cut almost always intersects
the medial wall border posterior to the olfactory sulcus. Also, when
drawing the registration borders, you don't need to rotate the surface
to approximate AC-PC alignment (e.g., when figuring out the starting
point for the aSTG landmark).
Since I only have one subject I can take
the time to do a better fit if a better fit is possible. Maybe Colin is
not the best approach (although it was tempting because I would like to
look at the AAL correspondence as well).
What is AAL?
Should I take a surface based
approach for this single subject to get the "best" fit. Thanks again for
your time.
What you're doing is definitely not the mainstream (inter-subject)
registration application, so in some ways I'm tempted to say go ahead
and use Colin. Most of the partitioning schemes available in the
PALS_B12 atlas were actually registered via colin, so they're more
"upstream" in the colin atlas. That said, the colin atlas doesn't have
the Amunt/Zilles cyto areas fMRI mapped, whereas PALS does (see PALS
tutorial -- Caret version, not webcaret at
http://sumsdb.wustl.edu:8081/sums/directory.do?dir_id=6332260&dir_name=PALS_ATLAS_TUTORIAL).
Also, David often fixes 'nits' he comes across in paint, border, and
coord files, and the PALS_B12 datasets get much more of his attention
than the colin atlas does.
BTW, the PALS paper hit NeuroImage early views. I don't have the link
handy but it is there as well.
Cheers,
-Morgan
On Tue, 2005-09-20 at 08:59 -0500, Donna Hanlon wrote:
Hi Morgan,
Many of our best users are FSL users, so we want to make life as easy as
possible for them. While we don't have Colin in FSL (what we're calling
"flirt") grid, we do have the PALS_B12 surfaces that way:
http://brainvis.wustl.edu/help/pals_volume_normalization/
We're recommending users move from Colin to PALS_B12 -- not only for
multi-fiducial mapping (MFM) of fMRI data, but *especially* for
registration purposes
(http://brainvis.wustl.edu/help/landmarks_core6/landmarks_core6.html).
See David's PALS paper for many important details:
http://brainmap.wustl.edu/resources/-PALS_ATLAS_June14_05.pdf
We're shooing people away from Colin and steering them to PALS_B12.
There might be some specific applications for which it makes sense to
use Colin, but I'd like to have a better understanding of them before
adding more Colin grids. Please elaborate a bit more on the requirement.
Space-dependent grids (more accurately in our context, volume
registration method dependent fiducial surfaces) generally have been
used to support fMRI mapping of group results to a visualization atlas
brain (e.g., Colin or PALS_B12). For registration, we never had
grid-dependent target datasets. This is because registration operates
on the sphere. It does matter what fiducial or average fiducial you use
when you visualize/project your registration results. Again -- there is
a flirt version of the PALS_B12 average fiducial in
CARET_HOME/data_files/fmri_mapping_files.
I realize this is probably somewhat confusing, so please let me know
where I need to be more clear.
Donna
On 09/20/2005 07:37 AM, Morgan Hough wrote:
I just wanted to nominate FSL output for grid support. I would be happy
to help supply any information but I believe you are already familiar
with the FSL tools. Please let me know if there is anything we can do to
make supporting FSL easier (such as for instance supplying Colin so that
people can register to that).
Thanks in advance for your time.
Cheers,
-Morgan
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
