OK, so Paraview screws up NIFTI coordinates, or doesn't read the header right?  
The surface mesh that ITK-SNAP makes is in vtk format, so it must be that 
ITK-SNAP doesn't put the orientation info into the right format?  

Looking closely at my image, I can see that the vtk mesh in Paraview is indeed 
left-right reversed compared to the original in ITK-SNAP (in addition to being 
shifted in various ways out of alignment inexplicably).  

Here are the bounds in Paraview for the mesh:

X range: -98.9 to -9.78 (delta: 89.1)

If I put the following for Reflect:
plane: X
center: 50

Then the images are no longer left-right reversed from each other, but they are 
still not overlapping.  There is some kind of Y and Z shift also going on.

-Tom


On May 17, 2012, at 11:51 AM, Philip A Cook wrote:

> Trying this out on some actual data - it seems the reflection should actually 
> be about the boundary of the image, not the center. Here the x-bounds of my 
> image are 114 to 337 in Paraview (note this does not agree with the NIFTI or 
> ITK coordinates), so I reflected the mesh (blue) about 114 in the x-axis, 
> giving me what I think is a correctly oriented mesh. 
> 
> Unfortunately, there appear to be other transformations at play as well. You 
> can't see it in this view but the red mesh is offset in the Z-axis. I think 
> this is because of the disagreement between the NIFTI and Paraview 
> coordinates.
> 
> <PastedGraphic-1.png>
> 
> On May 17, 2012, at 10:36 AM, Philip A Cook wrote:
> 
>> Hi,
>> 
>> I have had some success with the Reflect filter. Often the SNAP meshes are 
>> reflected about the origin of the image. Thus I do Reflect on the mesh (I 
>> always find it easier to transform the mesh rather than the image in 
>> Paraview), with Plane set to X / Y / Z as appropriate, and Center set to the 
>> Center of the image. 
>> 
>> The Center of the image in Paraview might not agree with what's in the image 
>> header, at least not for NIFTI files. You can figure out the Center in 
>> Paraview by selecting the image and then looking at the Information tab. The 
>> Bounds box will tell you the min and max in each coordinate. For the Reflect 
>> filter, you want the Center to be halfway between the min and max bound of 
>> the image.
>> 
>> This usually solves my problems with meshes and images not lining up. I'm 
>> assuming here that you are loading the same image from which the mesh was 
>> defined.
>> 
>> 
>> Phil
>> 
>> 
>> On May 16, 2012, at 11:53 PM, Tom Schoenemann wrote:
>> 
>>> OK, that worked. I've figured out how it works, following what you 
>>> suggested.
>>> 
>>> Unfortunately, Paraview didn't just shift the data over, leaving it in the 
>>> same orientation, it also inexplicably changed the orientation by several 
>>> degrees (??).  Even if I am able to get the two boxes in basically the same 
>>> location, the shifted orientation makes it impossible to be confident the 
>>> data are overlaid in the right places.  This is very frustrating.  
>>> Apparently Paraview makes random changes to the orientation??  Why on earth 
>>> would it do that?
>>> 
>>> -Tom
>>> 
>>> On May 16, 2012, at 5:18 PM, Scott, W Alan wrote:
>>> 
>>>> So, I just tried the following:
>>>> *Sources/ Wavelet.  Apply.  Display this by Outline.
>>>> * Filters/ Alphabetical/ Transform.  Translate X by 10.  Apply.
>>>> * Leaving the Transform filter highlighted in the pipeline browser, 
>>>> Filters/ Common/ Contor.  Apply.
>>>> * Now, if you turn on visibility of the original Wavelet, you see that the 
>>>> wavelet has moved along with the contour.
>>>> 
>>>> Or are you saying you want to move the two different files?  Then, you use 
>>>> two transforms – one per file.
>>>> 
>>>> Alan
>>>> 
>>>> From: [email protected] [mailto:[email protected]] 
>>>> On Behalf Of Tom Schoenemann
>>>> Sent: Wednesday, May 16, 2012 2:41 PM
>>>> To: Samuel Key
>>>> Cc: [email protected]
>>>> Subject: [EXTERNAL] Re: [Paraview] different orientation of input files
>>>> 
>>>> I must be missing something.  It only seems to move the bounding box, but 
>>>> not the location of the contour itself.. Not sure what that would be 
>>>> useful for, but in any case, I need to move the contour and/or data within 
>>>> the workspace, so the two overlap.  Transform is only changing the box, 
>>>> not changing the location of the object within the box.  Again, maybe I'm 
>>>> missing something.
>>>> 
>>>> -Tom
>>>> 
>>>> On May 16, 2012, at 2:32 PM, Samuel Key wrote:
>>>> 
>>>> 
>>>> Tom,
>>>> 
>>>> Will the Transform filter work for you? (Filters>Alphabetical>Transform)
>>>> 
>>>> It works for me with unstructured meshes. I don't know if it will work 
>>>> with a surface mesh produced ITK-SNAP.
>>>> 
>>>> Sam Key
>>>> 
>>>> On 5/16/2012 10:29 AM, Tom Schoenemann wrote:
>>>> Hi everyone,
>>>> 
>>>> I have a question about changing/editing the orientation and origin 
>>>> information about an input file.  Basically, the problem is this: I have a 
>>>> 3D volume (originally voxel data) file that contains statistics on its 
>>>> surface that I want to display color-coded.  I can create a nice smoothed 
>>>> version of the surface using a different piece of software (ITK-SNAP), and 
>>>> this is the surface I'd like to use to display my stats (Paraview does not 
>>>> allow you to smooth the surface in the same way that ITK-SNAP does, 
>>>> eliminating the stair-step surface artifacts).  I can indeed read both the 
>>>> stats file and the surface file from ITK-SNAP into Paraview, but 
>>>> unfortunately they now have different origins. That is, they don't overlap 
>>>> (the bounding box for the stats is completely outside the surface model 
>>>> that ITK-SNAP created.
>>>> 
>>>> Originally, both files had the same orientation and origin: ITK-SNAP used 
>>>> the same file to create the beautiful surface that the stats were created 
>>>> with.  
>>>> 
>>>> So Paraview apparently does not read the header info the same way for the 
>>>> two types of files, even though ITK-SNAP does.  
>>>> 
>>>> Is there a way to change the origin and/or orientation of files in 
>>>> Paraview?  I'm assuming there is a simple transform I could figure out by 
>>>> trial and error to get them to overlap again, but I don't know how or 
>>>> where I would do this.
>>>> 
>>>> Thanks for any suggestions!
>>>> 
>>>> -Tom
>>>> 
>>>> _________________________________________________
>>>> P. Thomas Schoenemann
>>>> 
>>>> Associate Professor
>>>> Department of Anthropology
>>>> Indiana University
>>>> Bloomington, IN  47405
>>>> Phone: 812-855-8800
>>>> E-mail: [email protected]
>>>> 
>>>> Open Research Scan Archive (ORSA) Co-Director
>>>> Consulting Scholar
>>>> Museum of Archaeology and Anthropology
>>>> University of Pennsylvania
>>>> 
>>>> Homepage: http://mypage.iu.edu/~toms/
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>> 
>>>> Visit other Kitware open-source projects at 
>>>> http://www.kitware.com/opensource/opensource.html
>>>> 
>>>> Please keep messages on-topic and check the ParaView Wiki at: 
>>>> http://paraview.org/Wiki/ParaView
>>>> 
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://www.paraview.org/mailman/listinfo/paraview
>>>> 
>>>> _________________________________________________
>>>> P. Thomas Schoenemann
>>>> 
>>>> Associate Professor
>>>> Department of Anthropology
>>>> Indiana University
>>>> Bloomington, IN  47405
>>>> Phone: 812-855-8800
>>>> E-mail: [email protected]
>>>> 
>>>> Open Research Scan Archive (ORSA) Co-Director
>>>> Consulting Scholar
>>>> Museum of Archaeology and Anthropology
>>>> University of Pennsylvania
>>>> 
>>>> Homepage: http://mypage.iu.edu/~toms/
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>> 
>>> _________________________________________________
>>> P. Thomas Schoenemann
>>> 
>>> Associate Professor
>>> Department of Anthropology
>>> Indiana University
>>> Bloomington, IN  47405
>>> Phone: 812-855-8800
>>> E-mail: [email protected]
>>> 
>>> Open Research Scan Archive (ORSA) Co-Director
>>> Consulting Scholar
>>> Museum of Archaeology and Anthropology
>>> University of Pennsylvania
>>> 
>>> Homepage: http://mypage.iu.edu/~toms/
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> Powered by www.kitware.com
>>> 
>>> Visit other Kitware open-source projects at 
>>> http://www.kitware.com/opensource/opensource.html
>>> 
>>> Please keep messages on-topic and check the ParaView Wiki at: 
>>> http://paraview.org/Wiki/ParaView
>>> 
>>> Follow this link to subscribe/unsubscribe:
>>> http://www.paraview.org/mailman/listinfo/paraview
>> 
>> _______________________________________________
>> Powered by www.kitware.com
>> 
>> Visit other Kitware open-source projects at 
>> http://www.kitware.com/opensource/opensource.html
>> 
>> Please keep messages on-topic and check the ParaView Wiki at: 
>> http://paraview.org/Wiki/ParaView
>> 
>> Follow this link to subscribe/unsubscribe:
>> http://www.paraview.org/mailman/listinfo/paraview
> 

_________________________________________________
P. Thomas Schoenemann

Associate Professor
Department of Anthropology
Indiana University
Bloomington, IN  47405
Phone: 812-855-8800
E-mail: [email protected]

Open Research Scan Archive (ORSA) Co-Director
Consulting Scholar
Museum of Archaeology and Anthropology
University of Pennsylvania

Homepage: http://mypage.iu.edu/~toms/









_______________________________________________
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Follow this link to subscribe/unsubscribe:
http://www.paraview.org/mailman/listinfo/paraview

Reply via email to