[caret-users] How to export flatten image as a 2D image?
Donna Dierker
donna at brainvis.wustl.edu
Fri Jul 17 09:12:40 CDT 2009
Hi Long,
This spec file has the AFNI fiducial coord file, specifically:
Human.PALS_B12.RIGHT_AVG_B1-12.FIDUCIAL_AFNI.clean.73730.coord
These are Talairach coordinates. There are several other versions of
the PALS fiducial coord file out there (and specifically in everyone's
caret/data_files/fmri_mapping_files subdirectory) that support
alternative stereotaxic spaces, like MNI152.
But if you are using a volume that comes with Mricron, then I'd be
surprised if it were in Talairach space; I'd expected MNI152.
The point is that you can't use a volume in MNI152 to sanity check the
alignment of data in Talairach space:
http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach
Find out more about ch2.nii. I wish I had time to, but I don't.
Donna
On 07/16/2009 09:40 AM, Long wrote:
> Hi Donna,
>
> It's really appreciate for your detailed answer. I am using the
> FIDUCIAL surface of PALS_B12.RIGHT.MCW_2008_06.73730.spec, and
> creating border files as you taught me. So, is that the stereoscopic
> coordinates which stored in the FIDUCIAL surface border file are the
> Talairach coordinates? I just simply treat them as the Talairach
> coordinates at beginning. If not, how can I convert these data to the
> Talairach coordinates?
>
> Best,
> Long
>
> On Thu, Jul 16, 2009 at 2:50 PM, Donna Dierker
> <donna at brainvis.wustl.edu <mailto:donna at brainvis.wustl.edu>> wrote:
>
> Hi Long,
>
> See inline reply below.
>
> Donna
>
> On 07/16/2009 04:45 AM, Long wrote:
> > Hi,
> >
> > Also, I try to merge my modified border data, which inside the BA 4,
> > on the FLATmap and FIDUCIAL map of
> > PALS_B12.RIGHT.MCW_2008_06.73730.spec, attached are the pics. I
> think
> > my question is due to the difference of BA temlate between
> Mricron and
> > Caret. Maybe I should trust one of them and I don't know which.
> >
> > Best,
> > Long
> >
> > On Wed, Jul 15, 2009 at 7:07 PM, Long <longspace82 at gmail.com
> <mailto:longspace82 at gmail.com>
> > <mailto:longspace82 at gmail.com <mailto:longspace82 at gmail.com>>>
> wrote:
> >
> > Many thanks to Donna and Alex! Now I begin to know the
> meaning of
> > border files and coord files.
> >
> > Recently, I tried to delineate the flat map with the border
> file.
> > I used PALS_B12.RIGHT.MCW_2008_06.73730.spec as a template.
> Then I
> > created a set of grid borders, which around the BA 4, and
> put the
> > FLAT border file into the Matlab. Then the flat map, which was
> > filled with dots, was delineated well in Matlab.
> >
> > I also export the border file of FIDUCIAL map.
> >
>
> Okay, the captures are very helpful in illustrating what you are doing
> -- thanks.
>
> > Then I used these coordinates data which were stored in that
> > border file to create a 3D matrix, and export this 3D matrix
> as a
> > nifti file.
> >
>
> Whoa. If you want to export something as NIfTI, which I think of
> as 3D
> voxel grid -- not sparse 3D points -- then you should use the BA4
> paint
> and use Attributes: Paint: Convert paint column to paint volume.
> You'll
> end up with a slab of voxels, and the "space" will depend on whatever
> fiducial surface you used. If you are using
> PALS_B12.RIGHT.MCW_2008_06.73730.spec, there is a good chance the
> default fiducial coord file is AFNI, which is considerably smaller
> than
> the MNI options (FLIRT, MRITOTAL, SPM*).
>
> I suppose if you do somehow use matlab to write the border points as
> NIfTI, then be mindful of which fiducial surface you use when
> saving the
> fiducial border file. Here are some useful documents:
>
> http://brainvis.wustl.edu/help/pals_volume_normalization/
> http://dx.doi.org/10.1016/j.neuroimage.2007.02.021
>
> > However, when I merge this nifti image on the ch2.hii.gz
> (provided
> > by Mricron), the voxels of this 'BA4 border' image were not
> > located around BA 4, and almost of them were located at BA 6. I
> > also try converting the coordinates data into MNI
> coordinates, but
> > the voxels still not around the BA 4.
> >
>
> I am unfamiliar with ch2.hii.gz. Hypothetically, let's say it is some
> population template in MNI152 space (may not be). Then try saving the
> border points (or Attributes: Paint: Convert paint column to paint
> volume) using the FLIRT or SPM* fiducials surfaces, and see if your
> results look more sane.
>
> >
> > Therefore, how can I create a correct 3D image with the border
> > file of FIDUCIAL map?
> >
> The procedure you are using to convert border points to voxels is a
> black box to me, so the problem could lie there. But it could be as
> simple as using the wrong reference fiducial surface when saving your
> fiducial borders.
>
> >
> > Thanks for your help.
> > Long
> >
> > On Thu, Jul 9, 2009 at 11:05 PM, Alexander Li Cohen
> > <alexc at npg.wustl.edu <mailto:alexc at npg.wustl.edu>
> <mailto:alexc at npg.wustl.edu <mailto:alexc at npg.wustl.edu>>> wrote:
> >
> > Yes,
> > as Donna stated, the 'generated Cartesian analysis
> grid' is the
> > Caret function I asked John to add so we could generate 2D
> > cartesian
> > grids on a flat surface, and then project the border so it's
> > folded
> > onto the fiducial surface... it IS a distorted grid (due
> to the
> > flattening), but dealing with a triangulated mesh in 2d, or
> > otherwise,
> > is much more difficult for me to visualize (not that it
> can't be
> > done... someday I hope to move towards that...) The
> subsequent
> > steps
> > are done in matlab and standard fMRI processing tools, with
> > occasional
> > use of caret_command to convert to and from surfaces if
> needed...
> >
> > transitioning to the sphere is in development, but of
> course,
> > it's a
> > lot harder and we're still working on it...
> >
> > -Alex
> > ~)---------------
> >
> >
> ----------------------------------------------------------------------
> > Alexander Li Cohen
> > alexc at npg.wustl.edu <mailto:alexc at npg.wustl.edu>
> <mailto:alexc at npg.wustl.edu <mailto:alexc at npg.wustl.edu>> (WORK Email)
> > alexcohen at gmail.com <mailto:alexcohen at gmail.com>
> <mailto:alexcohen at gmail.com <mailto:alexcohen at gmail.com>>
> (Non-secure /
> > rapid-response Email)
> > Petersen/Schlaggar Lab
> > Medical Scientist Training Program
> > Washington University in St. Louis School of Medicine
> >
> ----------------------------------------------------------------------
> >
> > This message is intended only for the exclusive use of the
> > intended
> > recipient(s) named herein and may contain information
> that is
> > PRIVILEGED and/or
> > CONFIDENTIAL. If you are not the intended recipient,
> you are
> > hereby
> > notified
> > that any use, dissemination, disclosure or copying of this
> > communication is
> > strictly prohibited. If you have received this
> communication in
> > error, please
> > destroy all copies of this message and its attachments and
> > notify us
> > immediately.
> >
> > The materials in this message are private and may contain
> > Protected
> > Healthcare Information or other information of a sensitive
> > nature. If
> > you are not the intended recipient, be advised that any
> > unauthorized
> > use, disclosure, copying or the taking of any action in
> > reliance on
> > the contents of this information is strictly prohibited. If
> > you have
> > received this email in error, please immediately notify the
> > sender via
> > telephone or return mail.
> >
> > On Jul 9, 2009, at 2:09 PM, Donna Dierker wrote:
> >
> > > Hi Long,
> > >
> > > Oh, okay. You mean this paper, where Alex detects areal
> > boundaries
> > > using functional connectivity:
> > >
> > > http://dx.doi.org/10.1016/j.neuroimage.2008.01.066
> > >
> > > This is a legitimate use of a flat map, although I think
> > that Alex
> > > Cohen
> > > may have figured out a way of using the spherical map,
> > instead. Over
> > > time, he has simplified this process. Alex is a
> caret-users
> > > subscriber,
> > > but I forward this reply to him, to make sure he sees it.
> > >
> > > The activation map for each subject is generated in
> volume-land.
> > > The 2D
> > > image analysis (edge detection, etc.) is better
> answered by Alex
> > > directly.
> > >
> > > But let me describe one way you could create the seeds:
> > >
> > > 1. Load flat map in main Caret window. Load fiducial
> > surface in
> > > Window 2.
> > >
> > > 2. Select Layers: Borders: Create Cartesian Grid Flat
> > borders and
> > > enter
> > > the desired spacing.
> > >
> > > 3. Select Layers: Borders: Project borders.
> > >
> > > 4. Select File: Save Data File: Border Files - Surface:
> > Associated
> > > surface -- change from Flat to Fiducial.
> > >
> > > The resulting border file can be opened in a text editor
> > and/or read
> > > in
> > > by a program like Matlab. It contains a bunch of border
> > points that
> > > represent this grid on the flat map projected onto the
> > fiducial --
> > > i.e.,
> > > the grid all crumpled up the way the cortex actually is.
> > Like I said,
> > > it is possible Alex is doing this on the PALS_B12
> sphere now
> > (using
> > > fiducial surfaces that have been resampled to the standard
> > mesh, but
> > > still look pretty much identical to the native mesh
> ones in
> > geometry).
> > >
> > > At any rate, there is no need to capture the flat map
> as an
> > image, nor
> > > any involvement with the Brodmann parcellation whatsoever.
> > On the
> > > contrary, if you do what Alex does, it is probably INSTEAD
> > OF Brodmann
> > > areas (although it would be fun to compare).
> > >
> > > Donna
> > >
> > > On 07/09/2009 09:35 AM, Long wrote:
> > >> Dear Donna,
> > >>
> > >> Thank you for your quickly reply. My main purpose is to
> > analyze the
> > >> cerebral cortex by using fMRI data in 2D way. Just like
> > your group
> > >> did
> > >> in paper: Defining functional areas in individual human
> > brains using
> > >> resting functional connectivity MRI. Neuroimage.
> 2008. How
> > to create
> > >> seeds and after calculating, how to create a 'activation
> > map' on a
> > >> flatten cortex and perform the 2D image analysis.
> > >>
> > >> Thanks for your help.
> > >> Long
> > >>
> > >> On Thu, Jul 9, 2009 at 3:05 PM, Donna Dierker
> > >> <donna at brainvis.wustl.edu
> <mailto:donna at brainvis.wustl.edu> <mailto:donna at brainvis.wustl.edu
> <mailto:donna at brainvis.wustl.edu>>
> > <mailto:donna at brainvis.wustl.edu
> <mailto:donna at brainvis.wustl.edu>
> > <mailto:donna at brainvis.wustl.edu
> <mailto:donna at brainvis.wustl.edu>>>> wrote:
> > >>
> > >> On 07/09/2009 06:30 AM, Long wrote:
> > >>
> > >> Dear caret-users,
> > >>
> > >> I want to analyze the flatten image as a 2D image,
> > so there
> > >> are a series of questions about this:
> > >>
> > >> 1) How to export a flatten cortex as a 2D
> image? So
> > I can use
> > >> some 2D image processing methods.
> > >>
> > >> I might misunderstand you, but I think it is as
> simple as
> > >> displaying the flat map in the main Caret window and
> > doing File:
> > >> Capture Image of Window.
> > >>
> > >>
> > >> 2) If a cortex was flattened, how to use the new
> > reference
> > >> coordinate system in 2D cortex to identify the 3D
> > coordinates,
> > >> e.g., MNI coordinates?
> > >>
> > >> In Caret, one typically would have a single topology
> > file for all
> > >> "CLOSED" surface configurations (e.g.,
> > fiducial/midthickness;
> > >> inflated; spherical), and a CUT topology file for any
> > flat ones.
> > >> The topology files store the neighbor relationships
> > between the
> > >> nodes, so you can see why a flat surface, which has
> > cuts, would
> > >> differ in its neighbors from a surface with no cuts,
> > like the
> > >> midthickness. But each of these different
> configurations --
> > >> fiducial, inflated, spherical, flat -- has its own
> > coordinate
> > >> (.coord) file. By default, Caret writes these coord
> > files as
> > >> binary; however, you can change the default to
> write them as
> > >> ASCII, so you can read in the coords into Matlab
> or whatever
> > >> software you like.
> > >>
> > >> If written in ASCII, the coord file will always
> have the
> > node
> > >> number in the first column, so you can grep or
> join the
> > 2D (flat
> > >> coord) and 3D (fiducial coord) files by the node
> number.
> > See
> > >>
> >
> http://brainvis.wustl.edu/CaretHelpAccount/caret5_help/file_formats/file_formats.html#coordFile
> > >> for more information about the coord file format.
> > >>
> > >> What is the relationship between 2D and 3D cortex.
> > >>
> > >> It is not a simple one, like an equation. Besides 2D
> > having cuts
> > >> (broken neighbor relationships), the nodes have been
> > projected
> > >> onto a Z=0 plane and then morphed across five
> cycles to
> > reduce
> > >> distortion and crossovers. But the node
> correspondence
> > remains
> > >> the same across configurations.
> > >>
> > >> It seems like you might want to do something like
> this:
> > >>
> > >> 1. click on the 2D image
> > >> 2. find the closest node on the 2D coord
> > >> 3. lookup up the corresponding 3D coordinate in the
> > fiducial coord
> > >>
> > >> This is easy in Caret, and if you have flat
> surface in main
> > >> window, fiducial in window 2, Caret will print the
> > fiducial 3D
> > >> coordinate in the ID window whenever you click on
> a node
> > in the
> > >> flat surface (main window).
> > >>
> > >> But once you export the flat to a 2D image, rather
> than
> > remain in
> > >> Caret, which is node-aware, you will have trouble
> > getting from
> > >> step 1 to step 2 above.
> > >>
> > >>
> > >> 3) How to export one or two flatten Brodmann areas
> > based on
> > >> the Brodmann template as a 2D image?
> > >>
> > >> This requires registering your individual subjects to
> > the PALS_B12
> > >> atlas, so you can deform the atlas Brodmann paint file
> > to your
> > >> subjects' flat maps. Alternatively, you can
> deform any
> > measure of
> > >> interest from your subjects to the PALS_B12 atlas, and
> > look at the
> > >> Brodmann regions on the atlas flat map, which has
> a standard
> > >> shape/outline (see attached capture).
> > >>
> > >> Be aware of distortions in the flat maps --
> compression
> > near the
> > >> center, expansion at the edges. We tend NOT to
> use them for
> > >> measurement/statistical purposes; rather, they are
> handy for
> > >> drawing borders and displaying an image of the entire
> > cortex in
> > >> one view. I wonder how you will be using them.
> > >>
> > >>
> > >> Thanks for your help.
> > >> Long
> > >>
> > >>
> >
> ------------------------------------------------------------------------
> > >>
> > >> _______________________________________________
> > >> caret-users mailing list
> > >> caret-users at brainvis.wustl.edu
> <mailto:caret-users at brainvis.wustl.edu>
> > <mailto:caret-users at brainvis.wustl.edu
> <mailto:caret-users at brainvis.wustl.edu>>
> > >> <mailto:caret-users at brainvis.wustl.edu
> <mailto:caret-users at brainvis.wustl.edu>
> > <mailto:caret-users at brainvis.wustl.edu
> <mailto:caret-users at brainvis.wustl.edu>>>
> > >>
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>
More information about the caret-users
mailing list