> Hmmm...
> I'm still a little unclear as to how the averaging works;
>     - The surfaces are created from 711-2C space.
>     - The volumetric SPM transform is applied to the 711-2C surfaces to
> place them in "SPM-surface" space
>     - The surfaces are averaged

Good so far.

> Does the variability map reflect the coord variability of the individual
> surfaces in 711-2C space, or SPM-surface space?

Currently, there is only a single 3d variability map and a single depth
variability map.  Both maps were generated using the 711-2B surfaces.

>>From my understanding, the surface average procedure doesn't impplement
>> any deformation to match the surfaces; it assumes the surfaces in are in
>> the same space. Thus, for example, it assumes that node i in persons j
>> and k should be in the same anatomical location.

Let's call it corresponding locations on the PALS standard mesh surface. 
These are not guaranteed to be the same anatomical location.  Figure 7 in
David's PALS paper shows how for a given node, some subjects vote with
sulcus A, whereas other subjects may vote gyral or sulcus B.

>> The only point at which
>> variability could be encoded (in my mind) would be in the normalization
>> from 711-2C to SPM space, but I can't see where this information is used
>> in the above processing steps (i.e., no transform file is used in the
>> above). So I'm guessing I'm missing something; either the averaging
>> procedure is more complex than I thought, or the transform info is stored
>> somewhere I'm unaware of (or I have it completely wrong!).

It's 9:41pm here in Milwaukee, and I've been working all day, so I
probably can't explain this as well as I'd like.  But it seems to me like
you're starting to grapple with what David calls "the correspondence
problem" -- which is a part of the learning curve of grasping exactly what
it means to register one surface to another on the sphere.

What I can assure you is that the step of creating the average coord file
is just a simple mean of the contributing coordinates -- no magic there.

The SPM *sn.mat (or something like that) does encode the nonlinear warping
transformation that gets applied to the surface coordinates.  If there is
a big difference between the 711-2B 3D variability and the SPM2 or SPM99
3D variability, then the 24 (12x2) *sn.mat files hold the key.
>
> Thanks again for your help,
> Alex
>
> PS - I cancelled the email with the pics. I can re-send it if you're still
> interested.

Not needed; I can look at this when I return to St. Louis, but it will be
a couple of weeks, because we're preparing for a Caret course in St.
Louis.  (I didn't advertise this on the list, because it's already pretty
full, and the facilities are less than optimal.)
>
> Alex Fornito
> M.Psych/PhD (clin. neuro.) candidate
> Melbourne Neuropsychiatry Centre and Department of Psychology
> National Neuroscience Facility
> The University of Melbourne
> Levels 2 & 3, Alan Gilbert Building
> 161 Barry St
> Carlton South Vic 3053 Australia
> Ph:    +61 3 8344 1624
> Fax:   +61 3 9348 0469
> email: [EMAIL PROTECTED]
>
>
>
> -----Original Message-----
> From: [EMAIL PROTECTED] on behalf of
> [EMAIL PROTECTED]
> Sent: Fri 5/26/2006 8:39 PM
> To: Caret, SureFit, and SuMS software users
> Subject: RE: [caret-users] Mapping foci
>
>>
>>
>> Ok, let me see if I got this right...
>>
>> 1 - I loaded the
>> Human.PALS_B12.LR.MULTI-FIDUCIAL_SPM2_fMRI-MAPPER.B1-12.LEFT.73730.spec
>> file from the caret/data_files/fmri_mapping directory and loaded all the
>> coord files
>
> Loading a spec file isn't necessary to create an average coord file.  (But
> it is generally helpful when visualizing the resulting average coord
> file.)
>
>> 2 - I created the average using Surface > Average coord file and
>> requested
>> the standard deviation surface_shape file.
>
> And you selected just the 12 left (or right) SPM2 coord files, excluding
> the average SPM2 coord file, right?
>>
>> 3 - I exited and reloaded the average coord file and surface_shape file.
>>
>> Compared to the 711-2C map, there seems to be a fair amount more
>> variability with SPM2 (I tried to attach pics, but the messgae was to
>> big
>> for the list).
>
> It's awaiting moderator approval; I'll get to that next.
>>
>> Just wandering why this would be the case. From what I understand, the
>> SPM2 normalization routine begins with an affine transformation,
>> followed
>> by a linear combination of basis functions to model nonlinear shape
>> differences. In contrast, it seems the 711-2C stops at the affine
>> registration. Could the greater variability in the SPM2 image result
>> from
>> the greater warping involved through the use of the additional basis
>> functions?
>
> I suspect the nonlinear component.
>
>> One other thing I'd like to clarify; are the surfaces for the SPM2 and
>> 711-2C PALS atlases created after the volumes were initially registered
>> to
>> their repsective templates, or was some
>> transform applied directly to the surfaces after they were reconstructed
>> in native space?
>
> Actually, the surfaces were reconstructed in 711-2C space.  The 711-2C
> volumes were input to the various spatial normalization methods (SPM99,
> SPM2, FLIRT, AFNI, and MRITOTAL), and the resulting volume transformation
> applied to the surface coordinates.  The link I provided in the earlier
> email provides details, but I don't have it handy here in Milwaukee
> (helping out with their fMRI/AFNI course).
>
>> Therefore, do the maps correspond to the SD of distance each voxel moves
>> during normalization, mapped onto the corresponding surface node, or the
>> distance moved at some other stage?
>
> No, I wouldn't characterize it this way.  It's not measuring movement
> during normalization.  Rather, I think of it as a measure of the spread of
> the cloud of points that represents the 3D coordinates for node i for each
> of the 12 PALS subjects.  One can compute the average coordinate for node
> i, and see how far each of the 12 subjects' coords stray from that
> average.  This is what the 3D variability maps measure.
>
> The depth variability maps, on the other hand, are a different animal.
>>
>>
>> Thanks again for your help!
>> Alex
>>
>>
>> Alex Fornito
>> M.Psych/PhD (clin. neuro.) candidate
>> Melbourne Neuropsychiatry Centre and Department of Psychology
>> National Neuroscience Facility
>> The University of Melbourne
>> Levels 2 & 3, Alan Gilbert Building
>> 161 Barry St
>> Carlton South Vic 3053 Australia
>> Ph:    +61 3 8344 1624
>> Fax:   +61 3 9348 0469
>> email: [EMAIL PROTECTED]
>>
>>
>>
>> -----Original Message-----
>> From: [EMAIL PROTECTED] on behalf of Donna Dierker
>> Sent: Thu 5/25/2006 11:29 PM
>> To: Caret, SureFit, and SuMS software users
>> Subject: Re: [caret-users] Mapping foci
>>
>> On 05/25/2006 03:27 AM, Fornito, Alexander wrote:
>>
>>>Never mind again!
>>>I found the surface_shape files in the tutorial, but there only seems to
>>> be the right hemisphere ones. I'm particularly interested in the 3D
>>> variability maps for the PALS atlas (left and right).
>>>
>>>
>> 3D variability -- not depth variability, right?  This shape file has
>> both left and right hem 3D variability:
>>
>> Human.PALS_B12.LR.B_1-12.BOTH-DEPTHnr_LoContrastAVG_StdDev_3D-Variability.73730.surface_shape
>> http://sumsdb.wustl.edu:8081/sums/archivelist.do?archive_id=6486399
>>
>> It has these columns:
>>
>> LoContrast AVERAGE B1-12 RIGHT DEPTH
>> 3D VARIABILITY - RIGHT_HEM Human.PALS_B12
>> LoContrast AVG DEPTH B1-12 LEFT (Women + Men)
>> 3D VARIABILITY - LEFT_HEM Human.PALS_B12
>>
>>>If you don't have the left hemispher map handy, is there anyway that I'd
>>> be able to generate one myself?
>>>
>>>
>> The 3D variability is generated using Surface: Create Average Coordinate
>> File, and checking the Create Sample Standard Deviation.  Just make sure
>> you choose the right coord files (e.g., don't mix left and right hems).
>> If you're averaging native coordinates, then expect much higher
>> deviation than if you're averaging AC-PC aligned, or better yet some
>> scale-controlled space (Talairach, icbm*, or 711-2*).
>>
>>>Also, does it matter what space the PALS atlas is in when viewing the
>>> variability map? There seems to be only one surface_shape file in the
>>> tutorial.
>>>
>>>
>> We only generated these measures on the 711-2C surfaces (the ones we
>> originally segmented).  I wouldn't expect significant differences using
>> the other spaces, although the SPM ones use nonlinear volume
>> registration methods, so those differences might be vary.  Try it and
>> let us know whether you think the differences are worth worrying about.
>>
>>>Sorry about all the emails anhd thanks again,
>>>Alex
>>>
>>>
>>>Alex Fornito
>>>M.Psych/PhD (clin. neuro.) candidate
>>>Melbourne Neuropsychiatry Centre and Department of Psychology
>>>National Neuroscience Facility
>>>The University of Melbourne
>>>Levels 2 & 3, Alan Gilbert Building
>>>161 Barry St
>>>Carlton South Vic 3053 Australia
>>>Ph:    +61 3 8344 1624
>>>Fax:   +61 3 9348 0469
>>>email: [EMAIL PROTECTED]
>>>
>>>
>>>
>>>-----Original Message-----
>>>From: Fornito, Alexander
>>>Sent: Thu 5/25/2006 1:10 PM
>>>To: Caret, SureFit, and SuMS software users; Caret, SureFit, and SuMS
>>> software users
>>>Subject: RE: [caret-users] Mapping foci
>>>
>>>Never mind, I found it. Its a separate coord file - right?
>>>Are there any surface-shape files for depth hidden anywhere, or do I
>>> generate that myself?
>>>Thanks again,
>>>Alex
>>>
>>>Alex Fornito
>>>M.Psych/PhD (clin. neuro.) candidate
>>>Melbourne Neuropsychiatry Centre and Department of Psychology
>>>National Neuroscience Facility
>>>The University of Melbourne
>>>Levels 2 & 3, Alan Gilbert Building
>>>161 Barry St
>>>Carlton South Vic 3053 Australia
>>>Ph:    +61 3 8344 1624
>>>Fax:   +61 3 9348 0469
>>>email: [EMAIL PROTECTED]
>>>
>>>
>>>
>>>-----Original Message-----
>>>From: [EMAIL PROTECTED] on behalf of Donna Dierker
>>>Sent: Tue 5/23/2006 12:45 AM
>>>To: Caret, SureFit, and SuMS software users
>>>Subject: Re: [caret-users] Mapping foci
>>>
>>>Hi Alex,
>>>
>>>Just a few things to add to John's replies:
>>>
>>>* Why are you using Colin instead of PALS?  Unless you have a very
>>>specific reason to use Colin, I would use the PALS_B12 atlas for this
>>>purpose.  In this case, you would use the average fiducial corresponding
>>>to the stereotaxic space.  You will find several of them in your
>>>$CARET_HOME/data_files/fmri_mapping_files subdirectory.  This page
>>>describes the methods used for normalizing the 12 Buckner brains to the
>>>various spaces:
>>>
>>>http://brainvis.wustl.edu/help/pals_volume_normalization/
>>>
>>>* The above link also explains how FLIRT, SPM2, and SPM99 use the same
>>>template/target, but the methods produce subtle differences
>>>(http://brainvis.wustl.edu/help/mni305_diffs/arch_diffs_mni305.html) --
>>>enough so that David felt it was worth having separate surfaces.  I
>>>tried to talk him out of it, mostly because I'm lazy.;-)  (Notice there
>>>is no SPM5 version yet.;-)
>>>
>>>* The AFNI coords are the only ones we have in "true" Talairach space,
>>>so yes -- these are the ones to use for mni2tal'd coordinates.
>>>
>>>* I know nothing about SPM96, but the spm_templates.man file in my SPM2
>>>distribution contains this helpful excerpt:
>>>
>>>% In SPM96, we released  a  single  subject  brain  for  use  as  a
>>>% template.   Although  the  MNI  gave  us  this  data,  they never
>>>% recommended that this brain  should  be  used  as  a  stereotaxic
>>>% standard.   This   is  something  that we at the FIL chose to do.
>>>% The official standard  for  the   ICBM   stereotactic  space   is
>>>% the   MNI305  brain  -  which  this was not.  This brain has many
>>>% merits for simulation but it  suffers   from   all   the   single
>>>% brain  criticisms  that apply to Talairach.  In this release, the
>>>% single subject T1 has been replaced by a 152 subject average.  We
>>>% (in  the  SPM  group)  chose  to  use the 152 subject T1-weighted
>>>% average rather than the 305 brain average because there are  also
>>>% T2-,  and  PD-weighted  images of the same subjects.  This should
>>>% allow  much  more  flexibility  in  the  range  of  different  MR
>>>% contrasts   that   can   be  spatially  normalized  to  the  same
>>>% stereotaxic space (by registering  to  a  linear  combination  of
>>>% template images).
>>>
>>>On 05/22/2006 08:55 AM, John Harwell wrote:
>>>
>>>
>>>
>>>>Alex,
>>>>
>>>>When mapping foci using the Map Stereotaxic Focus Dialog, the focus
>>>>entered is projected using the fiducial surface displayed in the main
>>>>window.  So, the main window surface must be in the same stereotaxic
>>>>space as the focus that is being mapped.  The stereotaxic space entry
>>>>on the Map Stereotaxic Focus dialog is a meta-data entry for now.
>>>>
>>>>Map foci as follows:
>>>>
>>>>1) As needed, change the fiducial surface in the main window so that
>>>>it is in the appropriate stereotaxic space for the focus currently
>>>>being mapped.  It is okay to load multiple fiducial surfaces in Caret.
>>>>2) Save the foci projection file.
>>>>3) Exit Caret.
>>>>4) Start Caret.
>>>>5) Reopen the spec file and choose the foci projection file just
>>>>saved and choose only one of the fiducial surfaces.
>>>>
>>>>Since the foci projection file stores the foci relative to a triangle
>>>>in the surface, the foci will be approximately transformed from the
>>>>other stereotaxic spaces.
>>>>
>>>>Good Luck.
>>>>
>>>>----------------------------------------------------------
>>>>John Harwell
>>>>[EMAIL PROTECTED]
>>>>314-362-3467
>>>>
>>>>Department of Anatomy and Neurobiology
>>>>Washington University School of Medicine
>>>>660 S. Euclid Ave.    Box 8108
>>>>St. Louis, MO 63110   USA
>>>>
>>>>On May 22, 2006, at 12:36 AM, Fornito, Alexander wrote:
>>>>
>>>>
>>>>
>>>>>Hi,
>>>>>I'd like to map some foci from past studies onto the Colin surface.
>>>>>I've noticed that there seems to be 2 options for selecting co-
>>>>>ordinate type during this process;
>>>>>
>>>>>1 -  when loading the Colin spec file, there is the option of
>>>>>choosing the SPM99, SPM2, etc fidcucial surface.
>>>>>2 - in the 'studies' tab of the 'map stereotaxic focus' dialog, it
>>>>>is possible to choose the stereotaxic space used in the study.
>>>>>
>>>>>Just want to check that I've got this right:
>>>>>
>>>>>If I chose the Colin fiducial to be SPM2, then that means the
>>>>>surface has been registered to SPM2 template space. If I want to  map
>>>>>the focus of a study that  used SPM99 space, will choosing  SPM99 in
>>>>>the 'map stereotaxic focus' dialog automatically convert  the coords
>>>>>so the focus appears in the (roughly) correct spot on  the SPM2-space
>>>>>surface?
>>>>>
>>>>>If so,if a study has reported coords that have been transformed  from
>>>>>MNI to talairach (e.g., using Matthew Brett's method), would
>>>>>choosing 'AFNI' in the 'map stereotaxic focus' dialog be the most
>>>>>appropriate?
>>>>>
>>>>>Finally, there are some studies that used SPM96. Does anyone know  if
>>>>>there are large differences between SPM96 and SPM99 spaces, or
>>>>>indeed, between SPM99 and SPM2 spaces?
>>>>>
>>>>>Thanks for your help,
>>>>>Alex
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>Alex Fornito
>>>>>M.Psych/PhD (clin. neuro.) candidate
>>>>>Melbourne Neuropsychiatry Centre and Department of Psychology
>>>>>National Neuroscience Facility
>>>>>The University of Melbourne
>>>>>Levels 2 & 3, Alan Gilbert Building
>>>>>161 Barry St
>>>>>Carlton South Vic 3053 Australia
>>>>>Ph:    +61 3 8344 1624
>>>>>Fax:   +61 3 9348 0469
>>>>>email: [EMAIL PROTECTED]
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>_______________________________________________
>>>>>caret-users mailing list
>>>>>[email protected]
>>>>>http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>>>>>
>>>>>
>>>>_______________________________________________
>>>>caret-users mailing list
>>>>[email protected]
>>>>http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>>
>>
>>
>> --
>> Donna L. Dierker
>> (Formerly Donna Hanlon; no change in marital status -- see
>> http://home.att.net/~donna.hanlon for details.)
>>
>> _______________________________________________
>> caret-users mailing list
>> [email protected]
>> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>>
>>
>>
>>
>>
>> _______________________________________________
>> caret-users mailing list
>> [email protected]
>> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>>
>
> _______________________________________________
> caret-users mailing list
> [email protected]
> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>
>
> _______________________________________________
> caret-users mailing list
> [email protected]
> http://pulvinar.wustl.edu/mailman/listinfo/caret-users
>

Reply via email to