[Insight-users] Clarification on itkOrientImageFilter orientationcodes

[Mark Wyszomierski]

I'm experiencing some problems in with image orientation translations
for axial images;

When translating a coronal image to sagittal, everything works as expected:
start w/: RSA
ask for:  ASL
get:       ASL

However, when translating an axial image to sagittal, I'm ending up
with a coronal image instead:
start w/: RAI (or LAS)
ask for:  ASL
get:       IRA (or SLA)

This result is confusing.  Essentially, the RSA=>ASL translation
requires a 90 degree turn clockwise along the y-axis, which itk does
correctly.  Meanwhile, the RAI=>ASL translation requires a 90 degree
turn clockwise around both the x-axis and y-axis.  But it seems for
RAI=>ASL, itk is instead rotating 90 degrees clockwise along the x-
and *z-axis* which results in a coronal image of IRA rather than a
sagittal of ASL...   any idea why this is happening?

In attempting to work around the problem, I pretended that my axial
image (RAI) was instead a coronal image (RSA).  I instructed itk to
translate the image to ARS (an unorthodox axial image) which
essentially is asking itk to rotate the image 90 degrees clockwise
around the x-axis and y-axis; the same as an RAI=>ASL translation
would ask.  This translation did indeed result in the expected output
of an ASL image.

Can anyone shed light onto the problems w/ translating axial to
sagittal?  thanks,

Mark

On 4/21/05, Miller, James V (Research) <millerjv at crd.ge.com> wrote:
> For DICOM files, you would need to calculate codes from the
> ImageOrientationPatient and not the ImagePositionPatient.
> 
> The codes in the OrientImageFilter label the "start" of the
> row, column, slices.  So in that nomenclature, your Desired
> orientation would probably be ASL.
> 
> Bill Lorensen has been working on some changes to the OrientImageFilter
> so that it will automatically set the Given orientation from the
> direction cosines of the image (ImageOrientationPatient), so the user
> would only have to specify the Desired orientation.  He is also
> adding convenience methods so that the user can simply request
> Axials, Coronals, or Sagittals instead of having to specify
> the codes.
> 
> It doesn't look as though his changes to the OrientImageFilter
> have been checked in yet.  But he is close to having it ready
> for committing to the repository.
> 
> Jim
> 
> 
> -----Original Message-----
> From: insight-users-bounces at itk.org
> [mailto:insight-users-bounces at itk.org]On Behalf Of Mark Wyszomierski
> Sent: Thursday, April 21, 2005 9:45 AM
> To: insight-users at itk.org
> Subject: [Insight-users] Clarification on itkOrientImageFilter
> orientationcodes
> 
> Hello all,
> 
> I implemented the itkOrientImageFilter so I can snap all my volumes to
> a particular orientation.The input/output is working, I am just unsure
> of what the orientation codes mean. For instance, I would like all my
> images to have their rows going towards the Posterior, columns going
> towards the Feet, and the Z (slice direction) going towards the Right
> side of the patient. How do I translate that into the codes provided
> by the filter?
> 
> Also, since my input are DICOM files, can I translate the
> ImagePositionPatient tag directly to one of the filter codes?
> 
> Thanks for any clues,
> Mark
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>