iv3d-users

Text archives Help


[IV3D-USERS] Re: ImageVis3D's coordinate system


Chronological Thread 
  • From: Dženan Zukić <dzenanz@gmail.com>
  • To: Jens Krueger <jens@sci.utah.edu>, ImageVis3D users mailing list <iv3d-users@sci.utah.edu>
  • Subject: [IV3D-USERS] Re: ImageVis3D's coordinate system
  • Date: Tue, 7 Dec 2010 21:53:59 +0100
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; b=FVojqYDNteWvlIKujz1i9MAmZnSYPAO0tb1srAYNf0knJLli5xa1GKrOBWq6dHunHs 0Fii0fGfE4uevas3e9UDqV1U+4Doyc1gsMjZ918+62YcOfqMUXKlLTCIhFe+uryy00ro y/FmwmN+E4h98+hhWdXJAg9sZMMeuibPkmbOk=
Spacing is, if I remember correctly, 0.625, 0.625, 4.4,  (x y z). If it is different, I will let you know tomorrow, when I get back to University. But there are two different numbers in these datasets I think, slice thickness (4mm) and slice spacing (4.4mm), so there is 0.4mm gap between slices. But lets not split hairs.
 
The spacing between slices in DICOM datasets should be determined from positions of origins of each slice in DICOM's world coordinate system. The other tags (slice spacing among them) are just a "convenience", and they can be wrong sometimes. I think there is a lot of information about it, http://www.itk.org/pipermail/insight-users/2008-November/027895.html or http://forum.dcmtk.org/viewtopic.php?p=7847&sid=d49544814de38f779af9635ed88a843d

>>... For geometry you can already specify any affine transformation...
When I say "I attempted", I meant I tried specifying NRRD's transformation matrix, inverse of that matrix (if you just use dataset's coordinate system this should have worked) and a few similar combinations. None of that worked, so that's why I asked what kind of coordinate system you use internally.

And yes, now that you support adding polygonal data you should think about adding matrix transformation (from NRRD for example) on top of the 3D scalar field data in order to have proper positions in "DICOM's world coordinate system" (world space of pretty much all medical data even in other formats such as NRRD), so that polygonal surfaces do not hover in the distance when they should be inside the 3D image. I can help you with this with some code and examples (tomorrow).

Regards,
Dženan

2010/12/7 Jens Krueger <jens@sci.utah.edu>
Just as a side note, are you aware that those DICOMs store a lot of private information about you 

e.g. 
name= Zukic Dzenan
birthday_30.10.1983
weight=90 kg
hospital=Kreisklinikum Siegen
name of pysician=Praxis Dr. Richter
patient ID= 0000307984
to name just a few

You may want to consider anonymizing them before you make them publicly available :-)

Gruß
 Jens Krüger

------------------------------------------------------------

Interactive Visualization and Data Analysis Group
http://www.ivda.uni-saarland.de

Prof. Dr. Jens Krüger
Research Group Leader, MMCI
Senior Researcher, DFKI
Adjunct Professor, SCI
Adjunct Professor, University of Utah

Campus  D3 4, Raum 2.32
66123 Saarbrücken

Telefon: +49 681 302-70750
Mobil: +49 1512 5377420

Am 07.12.2010 um 13:01 schrieb Dženan Zukić:

Hi guys,

what kind of coordinate system ImageVis3D uses?

Or more to the point, I have some polygonal segmentations of tumors in 3D images, and I want them to appear in their proper place in 3D space in iv3d. What I care about is relative position of polygonal model to the image (I want some nice visualizations). From my attempts to achieve this, I would say you are not using "World Space" as recorded into the images. It would be useful if you implemented world coordinate system, but if not (or meanwhile) I would like to know which transformation should I apply to my polygonal surface (gray in the attached image, with vertices V1..V5) in iv3d in order to have it relate properly to scalar image.

In the attached jpeg, 2D image is highlighted in red, world's coordinate system is black and dataset's coordinate system is blue. Both DICOM images and NRRD record this information into their files. By the way, here are some DICOM datasets which do not convert properly to UVF (scaling is wrong, i.e. Z-spacing): http://www.cg.informatik.uni-siegen.de/People/Zukic/data/Downloads/DzZ.zip

Regards,
Dženan
<DICOM_coordinate_system.jpg>



Archive powered by MHonArc 2.6.16.

Top of page