[Insight-users] how to use the new segmentation methods?

[Luis Ibanez]

Hi Orientation f,

I'm looking at Chunming Li's web page

          http://vuiis.vanderbilt.edu/~licm/

and trying to identify the specific publication that you mention.



Could you please specify which one of the following papers is
relevant:

    1. Chunming Li, Chiu-Yen Kao, John C. Gore, and Zhaohua Ding, 
"Minimization of Region-Scalable Fitting Energy for Image Segmentation", 
to appear in IEEE Trans. Image Processing.

    2. Chunming Li, Chenyang Xu, Changfeng Gui, and Martin D. Fox, 
"Regularized Level Set Evolution and its Application to Image 
Segmentation", submitted, 2007.

    3.  Chunming Li, Chiu-Yen Kao, John C. Gore, and Zhaohua Ding, 
"Implicit Active Contours Driven by Local Binary Fitting Energy", CVPR 
2007. (pdf, BibTex, demos)

    4.  Chunming Li, Chenyang Xu, Changfeng Gui, and  Martin D. Fox, 
"Level Set Evolution Without Reinitialization: A New Variational 
Formulation", CVPR 2005. (pdf, BibTex, demos)

    5. Chunming Li, Jundong Liu, and Martin D. Fox, "Segmentation of 
Edge Preserving Gradient Vector Flow: An Approach Toward Automatically 
Initializing and Splitting of Snakes", CVPR 2005. (pdf, BibTex, demos)

    6. Chunming Li, Jundong Liu, and Martin D. Fox, "Segmentation of 
External Force Field for Automatic Initialization and Splitting of 
Snakes", Pattern Recognition, 2005.


Should we look at any of the code in the following page:
http://www.engr.uconn.edu/~cmli/code/   ??


Please be more specific on the method that you are interested
on, so that we can contact Dr. Li with a *specific* request
for his code.


     Thanks,


        Luis


----------------------
orientation f wrote:
> Hi Luis Ibanez
> Thanks for your reply.
> I use the geometric active contour models to segment MR image of knee. 
> The edge of cartilage is blur and has high variations on its gradient 
> values. So the contour can't converge to the real boundaries exactly. 
> In recent year, Chunming Li, working at level set method, proposes some 
> new variational level set methods which do well in medical image 
> segmentation in a rather fast way. If these methods are ported to ITK 
> frame-work, fine result may be gotten.
> 
>  
> 2008/3/31, Luis Ibanez <luis.ibanez at kitware.com 
> <mailto:luis.ibanez at kitware.com>>:
> 
> 
>     Hi Orientation f
> 
> 
> 
>     1) Being "old" is not necessarily bad.
>        Algorithms do not have expiration dates.  :-)
> 
> 
> 
>     2) I haven't seen many algorithms getting "embarrassed",
>        this tend to be a rather exclusive human emotion.
> 
>        Have you found a particular instance of a segmentation
>        problem where you are having difficulty obtaining results
>        with existing ITK segmentation methods ?
> 
>        If so, could you please share with us the specifics of this
>        segmentation problem, and describe the details of the methods
>        that you have attempted, along with the specific parameters that
>        you have tested ?
> 
>        We prefer to discuss issues only on technical terms, and with
>        enough information for reproducing the problem that you may have
>        encountered.  Anecdotic reports with emotional overtones tend to
>        be rather useless, we delegate those to the realm of decadent
>        journals and conferences dedicated to vanity research, where
>        emotions such as "embarrassment" naturally belong.
> 
>        The goal of ITK is not to play the childish game of "we have the
>        best algorithms". Instead we focus on providing implementations
>        of useful algorithms using sound software engineering. When new
>        methods become available, we are more than happy to adopt them
>        and include them in the toolkit, as long as they are not infected
>        with intellectual "property" burdens, such as undisclosed patents.
> 
> 
> 
>     3) Does ITK gets updated ?
> 
>        Yes, it gets updated through contributions
>        to the Insight Journal.
>     http://www.insight-journal.org/InsightJournalManager/index.php
> 
>        These contributions are ported to the Insight/Code/Review
>        directory, where they are brought up to the software engineering
>        standards of the toolkit, and are refined to match the requirements
>        of our backward compatibility policy.
> 
>        Details of these procedure are described in the ITK Wiki:
>     http://www.itk.org/Wiki/ITK_Procedure_for_Contributing_New_Classes_and_Algorithms
> 
> 
> 
>     4) Where can you download new ITK files:
> 
>        a) From the CVS repository, every day.
>           Recently added algorithms/classes will
>           be found in the Code/Review directory.
> 
> 
>        b) From the Insight Journal, where contributions include source code,
>           image data and the parameters required to *reproduce* results.
> 
> 
> 
>     5) You seem to be interested in finding an open source implementation
>        of fast variational level set methods.
> 
>        Do you have a specific method in mind ?
>        Maybe a specific paper in mind ?
> 
>        If so, could you please let us know which one(s) ?
> 
>        We will happy to contact the authors to request them to make
>        available their source code, so that we can port it to the
>        Insight Toolkit and make it available to the larger medical
>        image analysis community.
> 
> 
> 
>     Please let us know,
> 
> 
>          Thanks
> 
> 
> 
>             Luis
> 
> 
> 
>     -----------------
>     orientation f wrote:
>      > hi,Luis Ibanez
>      > I use ITK for medical image segmentation. The methods provided in ITK
>      > are relative old, and become "embarrassed" when face complicated
>     images.
>      > A lot of new segmentation method, for example, fast variational
>      > level-set method, are proposed in recent years. Does the ITK updated
>      > with the time? If it does, where can I download the new files?
>      > thank you very much!
>      > best wishes!
> 
>