[Insight-users] itk + vtk segmentation questions
Luis Ibanez
luis.ibanez at kitware.com
Thu Jul 22 15:41:44 EDT 2004
Hi Jacob,
1) There are multiple methods for constructing 3D data
sets from a set of slices.
In the case of the Visible Human, the simplest
thing to do is to write a MetaImage header file.
You will find instructions on how to do this in
the Data link
http://www.itk.org/HTML/Data.htm
You will find one for the Visible Woman dataset in
the ftp data repository:
ftp://public.kitware.com/pub/itk/Data/VisibleWomanHead/RawRGB/
ftp://public.kitware.com/pub/itk/Data/VisibleWomanHead/RawRGB/VisibleWomanHead.mhd
1.5) The reason why you don't find many examples on 3D
segmentation in the Software Guide is because it
was harder to find a way of presenting the results.
However, almost all the segmentation methods in ITK
are implemented for N-Dimensions. From the point of
view of writing code, you can convert most of the
examples in the software guide from 2D to 3D just by
changing the template parameters of the Image type
instantiation.
You will find *many* examples in 3D segmentation in
the InsightApplications checkout.
In particular
- RegionGrowing
- FastMarchingLevelSet
- ShapeDetectionLeveSet
- GeodesicActiveContoursLeveSet
- ThresholdSegmentationLevelSetFltkGui
- Morphogenesis
- SegmentationEditorFltkGui
- DeformableModels
- SNAP (Level Sets)
- VolviewPlugins (many algorithms into VolView)
- LiverTumorSegmentation
Please look at the following page for description
http://www.itk.org/HTML/Applications.htm
The source code of InsightApplications can be downloaded
in the same way that you downloaded "Insight".
2) You can use VTK for visualizing the scans from the
Visible Human. In practice, once you load the data
set into an ITK image, then you can convert it into
a vtkImageData and visualize it with VTK.
You will find examples on this in
InsightApplications/Auxiliary/vtk
3) Are you using Shared Libraries ?
Did you built this example outside of the
Insight tree ?
What *exact* parameters did you passed in
the command line when the example crashed ?
Chances are that you are passing an inconsistent
number (or values) of parameters to this example.
Please post the exact command+arguments to the list.
Thanks
Luis
------------------------
Jacob Boomgaarden wrote:
> Hi all,
>
> I am working with a project to reconstruct a 3D model of the human lung
> from the Visible Human Project dataset and had a few questions regarding
> some segmentation usages. The application I will be making incorporates
> both ITK and VTK capabilities and, as of now, I am working with CT images
> while looking at using one or both of itkConfidenceConnectedImageFilter or
> itkFastMarchingImageFilter segmentation methods.
>
> Here are some questions I have come up against:
>
> 1) When using either of the above segmentation methods, how do I go about
> first creating a 3-dimensional itk::Image out of the 2D CT scans to then
> segment? I realize that the algorithms allow for 3D segmentation and I
> would assume that this would be more efficient to use than by segmenting
> each individual 2D image first and then creating a 3D image out of those.
> The examples that I have run across in the ITK SoftwareGuide seem to only
> show how to do the segmentation on a 2D image. Are there any examples
> using segmentation on 3D images?
>
> 2) Allow I haven't looked too hard within VTK's capabilities, I am just
> curious as to whether (and then how) VTK is able to view the bitmap images
> of the CT scans from the Visible Humag Project dataset?
>
> 3) While running the example
> Examples/Segmentation/ConfidenceConnected.cxx I ran into a peculiar
> problem. The first time I ran the executable it worked just fine and
> segmented the suggested BrainProtonDensitySlice.png image. But when I
> tried running it again at a later point, it wouldn't finish properly and
> displayed the following error message:
>
> Error: Segmentation Fault (core dumped)
>
> Why would this not work the second (and numerous tries afterwards) time
> correctly?
>
> Thanks for all your help and suggestions!
>
>
More information about the Insight-users
mailing list