[Insight-users] Suggestions and Help

Luis Ibanez luis.ibanez at kitware.com
Tue Jul 20 00:57:43 EDT 2004


Hi Jacob,

1) ITK provides methods that can do lung
    segmentation from CT datasets. Lungs
    presents as extremely dark regions in
    CT images. This makes them easy to
    segment out of the surrounding tissues.

    You can problably get an easy first
    segmentation by using the filter

     itkConfidenceConnectedImageFilter

    and then move to get a more elaborate
    segmentation by using the filter


      itkFastMarchingImageFilter


    Both filters are described in the
    segmentation chapter of the SoftwareGuide

    http://www.itk.org/ItkSoftwareGuide.pdf

    Your challenge will be to deal with
    the airways since the segmented regions
    will be connected to the traquea, the
    nasal cavity.... and then the external
    air.

    The solution is quite pragmatic, you
    can simply crop the dataset at the
    level of the neck and that will truncate
    the airways connection to the outside
    regions.

    That will give you a segmentation of the lungs
    represented in the form of an "image binary mask".

    You can take that binary mask, convert it to
    a VTK image and then use the contour filter
    in order to extract its surface.

    If you are not familiar with VTK you can
    easily experient with VolView and with ParaView.

    The free version of VolView can be downloaded
    from

     http://www.kitware.com/products/volview.html


    Source code and binary executables of ParaView
    can be downloaded from


          http://www.paraview.org



2) How to animate the lungs... that is a different
    story.... and it will certainly be more fun.   :-)


     It all depends on what physiological data you
     have for describing the respiratory motion....



     So...

           what kind of data do you have ?


    If you are also looking for data on this area,
    you may want to start by searching in Google
    for "respiratory motion".  You will find a lot
    of references on prediction and compensation
    of respiratory motion for radiation treatement
    and minimally invasive surgical interventions.



3) ITK does not performs Volume Rendering.
    Volume Rendering is a Visualization technique.
    ITK does not provide visualization functionalities

    You can do Volume Rendering with VTK
    You can do Volume Rendering with VolView.

    If you have limited time for performing such visualization,
    VolView is the way to go, since you can just use the
    free version for creating volume renderings interactively.



4) You can get more than a few slices from the Visible Human
    Dataset. You can get the *entire* dataset for *free* just
    by filling a form, signing it and returning it to the
    National Library of Medicine.

     http://www.nlm.nih.gov/research/visible/getting_data.html

  "A single License Agreement covering use of both the male and female
   Visible Human Project® datasets is available, as either a text file
   or a WordPerfect file. Please make two copies of the agreement and
   have both signed as originals by your appropriate officials. The
   agreement requires that you include a brief statement explaining your
   intended use of the dataset. Send both signed copies of the agreement
   and the statement of how you intend to use the data to:

     Visible Human Project®
     National Library of Medicine
     Building 38A, Room B1N-30
     8600 Rockville Pike
     Bethesda, MD 20894

   The agreement will be signed here at the NLM and one of the originals
   will be returned to you. Included with the signed agreement will be
   your account and password to the Visible Human Project® FTP site (if
   you wish to download all or part of the dataset via the internet),"



   You probably want to start simply with the CT dataset,
   and don't need to get it in full resolution.





Regards,



     Luis




-----------------------
Jacob Boomgaarden wrote:

> Hello All -
> 
> Thanks to all who have borne with me and continue to provide help to my
> que stions. Luis, your assistance with getting ITK to compile and build
> on my SGI was extremely helpful!
> 
> I'm doing a summer internship collaboration in Fairbanks, Alaska, at the
> Arctic Region Supercomputing Center and have not been receiving much
> direction or guidance from my mentor who is constantly out-of-touch with
> this project. His goal at the beginning of the summer was to do a 3D
> reconstruction of the human lung using the Visible Human dataset, and to
> then create an animation of this lung according to specific breathing
> patterns.
> 
> Doing our beginning research for this project led us in the direction of
> Kitware's open-source Toolkits (ITK and VTK). My specific portion of the
> project is that dealing with reconstruction of the lung from the dataset
> and I am having a hard time finding a first foothole to start with.
> 
> I've read many portions of the ITK software guide, as I believe the
> software system of ITK has the capabilities of what I want to achieve,
> but I am still somewhat fuzzy on which classes and algorithms I should
> be paying most attention to for my goals. With that in mind, I have some
> general questions I would like to pose:
> 
> 1.)  I am under the impression that the "Segmentation" algorithms that
> ITK provides are tools to accomplish what I am trying to do, but am
> unsure of this. After reading through examples it is still unclear to me
> whether these classes filter images creating a new image, or whether
> they segment regions and allow for outputing some sort of data
> representation of that region, or maybe I'm still not on the right
> understanding of these algorithms. Any information or strict capability
> definitions would be great.
> 
> 2.)  Some of the ideas that we have brainstormed about include reading
> in one image of data, extracting the contours or data points along the
> region of the lungs and repeating this idea to create some sort of
> structured point-set model or mesh of the lung. I am curious as to
> whether this is first of all, a feasible approach, and secondly whether
> ITK has the capabilities of doing this and if so, how?
> 
> 3.)  Another idea we had generated was doing something with the actual
> volume rendering of data over the lungs.Is this a better approach and is
> this at all feasible with ITK?
> 
> 4.)  Is there somewhere I can get a few sample data slices of the chest
> area of the Visible Human dataset that I can work with ITK on?
> 
> Any help and inspiration relating to my questions, or other issues
> pertaining to my project in general, that you can provide would be
> greatly appreciated, as I am unexperienced in this subject and haven't
> been able to receive much advice from my own advisor.
> 
> Thank you for your time in reading this and for all of your responses,
> 





More information about the Insight-users mailing list