[Insight-users] loading a dataset correctly

[Mark Wyszomierski]

Hi Dan,

I haven't used any of the itk classes for about a year - I'm just not
sure how the orientation class would be used here - I thought I need
to first load all my images into a 1D volume array, then apply the
filter to it? (if that's not right, please let me know). If it is,
here's my problem - say I have the following two studies:

instance #1 @ superior 50
instance #2 @ superior 10
instance #3 @ inferior -20

now I understand this means the scanner captured from head to foot.
Now I have this other case like:

instance #1 @ inferior -30
instance #2 @ inferior -15
instance #3 @ superior 18

and this should be captured foot to head. If I have to stuff these
images into a 1D volume array to be used with the ITK orientation
filter, at which 'slot' do I put instance #1 in the head to foot case,
and the same for the foot to head case?

Or is it setup to work so that I can just blindly load the images
ordered by instance number directly into the 1D array, give it the
actual orientation, and it will really be able to figure out the right
reshuffling?

I imagine I have to put them in the right order to begin with (ie. in
the head to foot case, instance #1 would go in the 'back', in the foot
to head case, instance #1 would go in the 'front').

Sorry for my confusion!

Thanks for your help,
Mark



On 3/8/07, Dan Mueller <d.mueller at qut.edu.au> wrote:
> Hi Mark,
>
> From my understanding the itk::OrientImageFilter determines the given
> orientation using the direction cosines (if UseImageDirection = true). I use
> Meta Header files with the AnatomicalOrientation field set to the
> orientation of the dataset on disk. Therefore when I read a dataset, I can
> use the itk::OrientImageFilter to set the desired orientation.
>
> In your case, some of your datasets will be "RAI" (ie. axial: right-to-left
> varies fastest, then anterior-to-posterior, then inferior-to-superior), and
> some of them "RAS". If you apply the itk::OrientImageFilter to the datasets
> and set the desired orientation "RAI", all the datasets will be transformed
> to be the same orientation.
>
> Is there some reason I missed preventing you from using this approach?
>
>
> Dan
>
> Mark Wyszomierski wrote:
> Hi Dan,
>
> I used this class about a year ago - if I remember correctly, it
> expects you to load the image data correctly in the first place before
> trying to transform.
>
> My problem is that for a give study, I don't know whether to load my
> images from back to front, or back to front.
>
> Do you know if that class has any support for this though?
>
> Thanks,
> Mark
>
> On 3/7/07, Dan Mueller <d.mueller at qut.edu.au> wrote:
> Hi Mark,
>
> You probably want the itk::OrientImageFilter
>
>
> http://www.itk.org/Doxygen/html/classitk_1_1OrientImageFilter.html
>
> HTH
>
> Dan
>
> Mark Wyszomierski wrote:
> Hi,
>
> I have two axial studies - one scanned from head to foot, the other
> scanned from foot to head.
>
> This causes a problem when loading them each a slice at a time into a
> vtkImageData object for volume rendering (based on instance number) -
> the acquisition scanned from head to foot comes out 'backwards'.
>
> I'm wondering if there is some ITK class to help me load the slices in
> the correct order?
>
> Please help me,
>
> Thanks,
> Mark
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>
>