[Insight-users] segmentation time!

[Luis Ibanez]

Hi Xujf,

1) Did you compiled your code for Release  ?

   In ITK that can easily change performance
   by a factor or 10X.


2) NarrowBand... as any other LevelSet method
   is an iterative method. You run it for as
   many iterations as you want.  You selected
   to run it for 800 iterations!!, The algorithm
   will stop when the RMS error is lower than
   the value that you provided or when it reaches
   the maximum number of iterations. What RMS
   value did you selected ?

   If you want to do a segmentation fast with
   LevelSets you must provide a good initial
   level set. Just starting with a blob may
   not be a good idea since it will take a lot
   of iterations for the surface of your blob
   to propagate up to the surface of your
   object.

   What are you providing as initial level set ?


3) You could also use a Multi-Resolution approach
   for making the LevelSet compute faster in lower
   resolution images.


There are *many* parameters involved in a segmentation,
specially when using LevelSets.


Reporting execution time for an algorithm without
describing the conditions in which you ran this
algorithm is meaningless. This is something that
you should only do when publishing papers and want
to claim that algorithm A is better than algorithm B.  :-)



Regards,


   Luis


--------------
xujf wrote:

> Hello,
>     Different segmentation approaches cost different time. my data is (180,260,180). 3 monthes ago,I realized the fast marching segmentation ,and it cost me about 14 seconds.these days , I use itknarrowbandcurvesimagefilter to segment the same data and it cost me more than fifty minutes. I wonder whether the narrow band segmentation cost so much time.
> 
> Best Regards!
> xujf 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users