[Insight-developers] two-input filters don't work

[Michael Xanadu]

Hi Kevin,

thank you for your reply.
I looked at my feature image and initial level set image and found out, that
both images suit your requirements. Actually, all level set filters work
fine with image files, like it's described in the manual. But when I use an
itk::GDCMImageIO as input - forget about it...
The clue is: Fast Marching Segmentation is the only one, which works with
dicom, the other three level set filters (Shape detection, Geodesic active
contours, Threshold Level Set) lead to the white spot I decribed in the last
mail. That led me to believe that the problem has something to do with the
"two-input" filters.
Do you have any further information for me how to solve the problem? If not,
I think I have to learn the hard way, like you. :-(

Regards, Michael


2009/9/4 Kevin H. Hobbs <hobbsk at ohiou.edu>

> On Thu, 2009-08-27 at 12:55 +0200, Michael Xanadu wrote:
> > Hi folks,
> >
> > I've got a problem with all segmentation filters which have two inputs
> > (normal input & feature image), for example ShapeDetectionLevelSet or
> > GeodesicActiveContour. I tryed the accordant examples in the example
> > folders and they work great (with image files). But if I replace the
> > input by a dicomreader (of course, I changed all accordant parameters,
> > too) I always get a black output with a little white spot at the
> > coordinate where I placed the seedpoint. I messed around with the
> > filter parameters, but I never get a segment, only the spot. Is
> > somebody familiar with the problem?
> >
> > Michael
> >
> > P.S.: dicom data = 512 x 512 x 4, short
>
>
> My guess is that your input images are not exactly right for these
> filters.
>
> The examples given in ITK take great pains to set up the feature image
> and the initial levelset "just so".
>
> The level set filters don't ever take a "normal input". I had to learn
> this the hard way (very hard way in fact I just discovered that some of
> my old work had edge images the wrong way around. It's no wonder they
> seemed to stop early!) by disassembling the examples into many
> reader->filter->writer programs and carefully looking at the outputs.
>
> The feature image should range from 0 to 1 with values close to one
> inside and outside the object, and values close to zero at its edges.
>
> The initial level set image should have negative values inside the
> object and positive values outside the object.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20090908/35f4d7d9/attachment.htm>