[Insight-users] which filter can I use?

[Luis Ibanez]

Hi Anne-Laure

The first thing that comes to mind is to use a region
growing filter, and placing its seed in one corner of
the image, in order to fill-in the background with a
different color.

However, presumably the lungs are connected (form a
single connected component) to the background through
the airways, so that region growing filter will probably
fill-in the lungs too.

You could fix this in two ways:


1) you could edit the image and "draw" in the nostrils
    in order to disconnect the lungs+airways, then run
    the region growing filter.

or

2) you could use the ThresholdLevelSet filter that
    behaves very close to a region growing, but has a
    curvature constrain. You could play with the curvature
    until you find a value that doesn't let the level set
    front enter the nostrils.


BTW, maybe I'm making up all this problem, because in a
typical lung scan, the head shouldn't show up in the field
of view. Isn't it ?  So you probably have a scan that only
contains the chest and part of the neck.

If that's the case, then the ConnectedThreshold filter
http://www.itk.org/Insight/Doxygen/html/classitk_1_1ConnectedThresholdImageFilter.html
should do the trick.


Just put seed points in the outside of the patient,
and select as "Output value" the gray level that you
want to use.


Yet another option is to simply try the ConnectedComponents
filter (assuming again that the scan finishes in the neck
http://www.itk.org/Insight/Doxygen/html/classitk_1_1ConnectedComponentImageFilter.html

Then follow it with the Relabel components image filter
http://www.itk.org/Insight/Doxygen/html/classitk_1_1RelabelComponentImageFilter.html


    Please let us know if you have additional questions,


       Thanks


           Luis



----------------
annelaure wrote:
> Hi,
> I have a problem and I don't know which filter or segmentation I can use. 
> My problem :
> I have a 3D scan of lungs and thorax. I made a binary image with the
> background of my scan equal to 0, the skin+muscles+fatty tissue equal to 255
> and lungs equal to 0. I would like to transform the background of my image
> in another color in order to differentiate the lungs and the background. Is
> it possible with ITK?
> 
> Thank you so much,
> Best regards,
> Anne-Laure