[Insight-users] Re: Shrink Level Set

[Luis Ibanez]

Hi XueQuan,


1) Please don't confuse the implementation of the filters with
    the examples of the filters.

    The Example of the GeodesicActiveContours uses seed points
    to feed a *FastMarching* filter that provides an initial
    level set for the GeodesicActiveContours filter.

    The GeodesicActiveContours filter itself *do not* take seed
    points as input.


2) Level Sets will alwasy produce segmentations that are a bit
    before the actual edge of the object.

    In the case of the GeodesicActiveContours filter you can
    try to get closer to the border by increasing the weight
    of the Advection component. Note that for the weight to have
    a noticeable effect you have to make it 10x or 100x larger
    than the weight used for propagation or curvature.

    Note also that the Advection term is only useful when your
    levelset contour is *already* very close to the real boundary
    of the object that you are segmenting.



Regards,



      Luis



==========================
qxue at jouy.inra.fr wrote:
> Hi, Lius:
> 
>    I have completed the all modules with your help. Thanks very much.
> 
>    Using the Geodesic Active Contour filter, I can get a region segmentation
> based on masked feature image. The result is not very satisfied, for I can not
> get the very exact segmentation, which is always smaller than the original
> cell. Which parameter should be changed to this question?
> 
>     So I want to try the shrink from the outside to inside.
> 
>     But I can not understand what you said, that is, "If you negate the initial
> level set, then you can have it behaving as a shrinking front instead of an
> expanding front".
> 
>    In the ITK book, one seed or a set of seeds are needed as initial condition.
> What should I do for a shrinking front except the negate the initial level set?
> Does ITK support edge instead of seeds?
> 
>    Best regards
> 
> XueQuan
> 
> 
> 
> Selon Luis Ibanez <luis.ibanez at kitware.com>:
> 
> 
>>Hi XueQuan,
>>
>>Here are some options that you may want to consider:
>>
>>1) You can take the Voronoi map image produced by
>>    the Danielsson or the Maurer distance map filters.
>>
>>2) For every marker, theshold the voronoi map, so that
>>    you end up with a mask that has the voronoi region
>>    corresponding to a single nucleus.
>>
>>3) Use this binary region to mask the speed image that you
>>    compute from the gradient magnitude of the original image.
>>    Pass the resulting masked image as the feature image to
>>    the level set. In this way you are restricting the level
>>    set to grow into the voronoi region at most.
>>
>>
>>
>>About your other question,
>>
>>If you negate the initial level set, then you can have
>>it behaving as a shrinking front instead of an expanding
>>front.
>>
>>
>>   Regards,
>>
>>
>>
>>       Luis
>>
>>
>>
>>-------------------------
>>qxue at jouy.inra.fr wrote:
>>
>>>Hello, luis:
>>>    I have sent an email to the insight-user, but i am not sure whether my
>>>question is clear. So I send you this letter again.
>>>    I want to apply the level set method similar to the attachment file.
>>>    Through DanielssonDistanceMapImageFilter, I have gotten the Voronoi
>>>segmentation image, but I can not use it to limit the region of level set
>>>method, for I do not know how to get the area of each Voronoi region.
>>>    May I get the edges of the every Voronoi segmentaion region getten
>>
>>directly
>>
>>>from the DanielssonDistanceMapImageFilter.
>>>    For I want to use the level set method to segmenation nuclei just in
>>
>>its
>>
>>>Voronoi region exactly.
>>>    The other question is how to apply a level set method in ITK from
>>
>>outside to
>>
>>>inside, that is, instead of one or more seeds, may I use the boundary of
>>
>>Voronoi
>>
>>>and set the velocity is negtive and then the level set is constracted from
>>>outside to inside, and then stops at the boundary of the nuclei.
>>>    Thanks,
>>>XueQuan
>>
> 
> 
>