Hi again Julian,

Sorry for the long-winded answer that follows, but these questions are worth taking some time to address.

In CARET_HOME/data_files/fmri_mapping_files (where CARET_HOME is the caret distribution directory), you will find numerous "fiducial" surface coord files in SPM2 space:


By "fiducial," I mean a surface that represents layer 4 of the cortex, or midway between the pial and GM/WM (Freesurfer white) surfaces. We also use the term "midthickness" in lieu of fiducial. David prefers the fiducial, because it neither over- or under-estimates the surface area of gyral crowns or sulcal fundi. (Pial surfaces underestimate the area of sulcal fundi, while white surfaces underestimate the area of gyral crowns.) In short, fiducial surfaces are just right. While these distinctions are less important for visualization purposes, they become very important for surface-based analyses (e.g., folding differences across groups).

The coord files named like Human_Buck_Case*.L.*coord are the left hemispheres of the 12 subjects included in our PALS_B12 atlas; the Human_Buck_Case*.L.*coord ones are the corresponding right hems. Any of these surfaces will appear more pleasing to someone who expects to see a pial surface than, say, an inflated surface. But it will still look kind of skinny to him/her.

Before David created the PALS_B12 atlas, we used the Colin atlas, and in fact the last three files listed above are the colin cerebellum, left hem, and right hem fiducials. But Colin's brain, like any single subject's, had quirks that made it unrepresentative of a population. For example, his forked STS caused activity to sometimes split across the posterior STS (see attached capture). Choosing Buckner 1 instead of Colin really isn't any better: This is just the sort of thing that happens with a single subject at.as. While David's PALS paper is a long, tough read, it's worth it: http://brainvis.wustl.edu/resources/-Pals.wcover.pdf

Freesurfer does generate pial surfaces. (And one can average the pial and white to generate a midthickness surface, which we encourage.) And I'm guessing you could probably use the "bert" pial surface that either comes with the distribution, or is easily downloaded. But we again gently nudge you away from the single subject representation toward a multi-subject atlas.

The Human.PALS_B12.*_AVG_B1-12*coord files are the PALS_B12 average fiducial surfaces, which are the best representation we have of normal group midthickness. (We also have counterparts for some proband groups, e.g., Williams Syndrome, autism, schizophrenia that are in varying stages of availability.) They're not as pretty as fiducial surfaces, but they're MUCH more representative. And if you want pretty, it's tough to beat the inflated surface with the sulcal depth underlay (i.e., the ready-made scenes in Human.PALS_B12.BOTH.TEMPLATE-for-fMRI-MAPPING.73730.spec).

Since you're using Human.PALS_B12.BOTH.TEMPLATE-for-fMRI-MAPPING.73730.spec, you already know it's possible to view both hemispheres in Caret concurrently, but each hem is in a different window. (For more details, take the full September 2006 tutorial http://sumsdb.wustl.edu/sums/directory.do?id=6585200.) I don't know of any plans to "un-split" the hemispheres (i.e., combine them into a single coord file). David seems to think of the hemispheres as mostly independent entities.

I'm sure you are not alone in your questions/concerns.


On 03/07/2007 04:39 PM, [EMAIL PROTECTED] wrote:
Thanks Donna, everything seems to work now, so I’m happy :)

About the surfaces, I saw something in the archive, is it a white brain,
or deflated pial:
and then click the first option with a brain icon on it that is possible
to view in the web browser.

There are 12 versions of each hemisphere so I presume that it is an
individual brain. Do you think is ok to use one of them, or do you happen
to have an average surface?

I think you are right, inflated brains may be best suited for visualizing
purposes, however, just in case some stubborn SPM guy would like once in a
while see he’s activation the way he is used to….
I think Freesurfer has a pial surface. Maybe it works to import that somehow?

Also, it would be nice to have the option to see the whole brain at once
(the non-split way) even inflated, but I guess that’s not yet possible….


On 03/07/2007 10:27 AM, Julianm Macoveanu wrote:
Human.PALS_B12.BOTH.TEMPLATE-for-fMRI-MAPPING.73730.spec file.

I think I got the hang of it. The break through was, after creating the
metric file load one of the predefined scenes before trying to open my
metric file. In this way it didn't complain anymore about different
of nodes.

If I understand it right if I want now to see the activation on a pial
surface I should just open the coord file and add it to my template spec
file. Do you know where I could find this? I think it should also be
something between pial and fiducial.

The PALS_B12 atlas (and the colin atlas before it) has no pial
surfaces.  One can generate pseudo-pial surfaces by using Surface:
Geometry: Expand or Shrink surface to expand a fiducial surface, but
I've never tried this with an average fiducial surface, and I don't
really recommend it.  We prefer the inflated surface for visualization,
unless you have activity in the operculum or insula that is difficult to
see; in such cases, we then use the Very Inflated surface.  Activity in
buried sulci isn't visible when viewed on a pial or a regular fiducial
(less so with an average fiducial).

I realize some people still prefer looking at a highly structured pial
surface, but considering this is group data you're mapping, the
structure of any one individual is not really representative of the
"group" structure of your subjects.  We believe mapping to PALS_B12 and
viewing the results on the inflated surface is a more faithful
representation of your results than viewing them on, say, colin27's pial
surface.  Slowly, we're hoping to win you over. ;-)

Also, L and R are overlaid wrong, what SPM reports as left, I see it on
right hemisphere in Caret. How can this be changed?

Use File: Open Data File: volume functional file and check the Volume is
an SPM2 Volume with Right on Left checkbox.

-----Original Message-----
[mailto:[EMAIL PROTECTED] On Behalf Of Donna
Sent: Wednesday, March 07, 2007 4:05 PM
To: Caret, SureFit, and SuMS software users
Subject: [BULK] Re: [caret-users] Map SPM volume
Importance: Low

Hi Julian,

When mapping to atlas, are you selecting a spec file created from this


... using steps as described in Section 5.1 of this tutorial:


It's tough to see how you could get this error under those
circumstances, since the files in that spec file (or derivative thereof)
all have the same 73730 standard mesh.

While this spec won't have a pial surface, it will have left and right
versions of the following configurations:

PALS_B12 average fiducial
very inflated

Since you're using SPM2, I'd suggest replacing the average fiducial
coord files with these files:


To do so, after opening
Human.PALS_B12.BOTH.TEMPLATE-for-fMRI-MAPPING.73730.spec, select File:
Open Data File: and select
Human.PALS_B12.LEFT_AVG_B1-12.FIDUCIAL_SPM2.clean.73730.coord.  Repeat
for Human.PALS_B12.RIGHT_AVG_B1-12.FIDUCIAL_SPM2.clean.73730.coord.
Then, select Toolbar: Spec and use the X buttons to delete these
versions, which are similar but not as optimal as the SPM2 versions in
your case:


Both the SPM2 and FLIRT versions are in mni152 space, but the
contributing PALS_B12 anatomicals were normalized using SPM2 for the
SPM2 versions, while FSL's flirt (affine only) was used to normalize the
FLIRT versions.


On 03/07/2007 08:33 AM, Julianm Macoveanu wrote:

Hi guys,

I need some help with basic stuff. I just want to use Caret to
visualize my SPM2 activations on inflated/Fiducial surfaces. So I have
the normalized volume file to start with.

I have opened a spec file from the Caret tutorial, use the map volume
to surface function, choose my volume, then Map to spec file with
atlas (and without). I have also tried not to open any spec file to
begin with, and then just open the one I chose as output during the
mapping procedure

As space I tried the SPM2, as this is the kind of volume I have (but
also 711-sc).

After the metric file is generated, I just try to open it, but I get
the error that my metric file contains different number of nodes.

What I would like to have is a spec file that have all different
surfaces (including pial and multi-fiducial which I could not find in
any examples of spec files) and then just create metric files from my
SPM volumes in order to visualize the activations.



caret-users mailing list

caret-users mailing list

Donna L. Dierker
(Formerly Donna Hanlon; no change in marital status -- see
http://home.att.net/~donna.hanlon for details.)

caret-users mailing list

caret-users mailing list

Donna L. Dierker
(Formerly Donna Hanlon; no change in marital status -- see 
http://home.att.net/~donna.hanlon for details.)

<<inline: colin.L.sts.jpg>>

Reply via email to