[Insight-users] Output of the atlas segmentation.

Lydia Ng lng at insightful . com
Thu, 21 Aug 2003 10:24:19 -0700 

Yasser,

The original purpose of the application was to run a validation study
using IBSR data. As such the only output that was needed was a list over
overlap measures and hence it does not output any images at all.

If you want to output images, you will have to modify your code locally.
Sample code for warping images and to output images can be found in the
SimpleAppOutput class and
Examples/Registration/DeformableRegistration2.cxx

- Lydia

> -----Original Message-----
> From: yasser salman [mailto:yass71 at yahoo . com]
> Sent: Tuesday, August 19, 2003 10:21 AM
> To: Lydia Ng; Luis Ibanez
> Cc: insight-users at public . kitware . com
> Subject: RE: [Insight-users] Output of the atlas segmentation.
>=20
> hi lydia..
> thanx for these details.., one more question plz..,
> is the output of the atlas_segmentation application is
> volume of images type PNG. i see in file
> SimpleAppOutput.txx use PGMVolumeWriter..
> thanx..
> yasser..
>=20
> --- Lydia Ng <lng at insightful . com> wrote:
> > >> can i use these data for  my own atlas-based
> > > > segmentation?
> >
> > Just to clarify:
> > Do you mean if you could use one of the IBSR data as
> > the moving (atlas)
> > and another image from another all together
> > different data set as the
> > fixed (to be segmented) image?
> >
> > If so, there are two issues you should be aware of:
> > intensity matching
> > and field of view.
> >
> > Demons is basically "mean squares" type approach.
> > That is it assume that
> > pixel belonging to the same structure has the same
> > intensity on both the
> > moving and fixed images. So if you are using images
> > from different
> > scanners with different settings and protocol you
> > will mostly likely
> > need to preprocess the images so that the intensity
> > matches.
> >
> > In the application there is some preprocessing done
> > with a
> > HistogramMatchingImageFilter which tries to match
> > intensity by matching
> > the image histograms at specified number of points.
> > If you want to use images from different data sets,
> > I strongly suggest
> > that you read the reference for the histogram
> > matching and know its
> > limitations.
> >
> > The other issue is the field of view, the classic
> > demons algorithm
> > assume that the structure in one image can be found
> > in the other image.
> > So if your two dataset have different field of views
> > then the algorithm
> > may also run into problems.
> >
> > - Lydia
> >
> >
> > > -----Original Message-----
> > > From: Luis Ibanez [mailto:luis . ibanez at kitware . com]
> > > Sent: Monday, August 18, 2003 12:03 PM
> > > To: yasser salman
> > > Cc: insight-users at public . kitware . com
> > > Subject: Re: [Insight-users] Output of the atlas
> > segmentation.
> > >
> > >
> > > Yasser,
> > >
> > >
> > > The fixed and moving image should have the same
> > dimensions,
> > > but not necessarily the same size. That is, both
> > images
> > > should be 2D, or both should be 3D. But the actual
> > number
> > > of pixels along each dimension may be different
> > between the
> > > two images.
> > >
> > > You should be ok using any atlas for your
> > segmentation,
> > > via demons deformable registration.
> > >
> > >
> > > Regards,
> > >
> > >
> > >    Luis
> > >
> > >
> > > ---------------------
> > > yasser salman wrote:
> > > > hi luis ..,
> > > > yes u understand acutally what i mean..but how
> > about
> > > > the data??  i know in the DemonsRegistration the
> > fixed
> > > > images and moving image must have the same
> > dimension
> > > > ..,SO i have a Volume downloaded from IBSR
> > > > (20Normal_T1 & 20Normal_T1_brain) which used in
> > the
> > > > atlas segmentaion application (IBSRValidation
> > study in
> > > > ITKApplication)
> > > > can i use these data for  my own atlas-based
> > > > segmentation? also i have a data from brain web
> > > > (Simulated Brain Database)  found in
> > > > www.bic.mni.mcgill.ca/brainweb.. how can i found
> > the
> > > > atlas to simulated brain Volume..? thanx Luis
> > for ur
> > > > helpfull support..
> > > > yasser..
> > > >
> > > >
> > > >
> > > > --- Luis Ibanez <luis . ibanez at kitware . com> wrote:
> > > >
> > > >>Hi Yasser,
> > > >>
> > > >>
> > > >>I'm not sure I understand your question...
> > > >>It seems that you are interested in setting up
> > > >>your own atlas-based segmentation procedure.
> > > >>
> > > >>
> > > >>The essential class of the segmentation is the
> > > >>DemonsRegistrationFilter
> > > >>
> > > >
> > > >
> > >
> >
>
http://www . itk . org/Insight/Doxygen/html/classitk_1_1DemonsRegistrationFi
> > lt
> > > er.html
> > > >
> > > >>You may want to start by looking at the
> > description
> > > >>that Lydia made of this filter on the
> > SoftwareGuide
> > > >>
> > > >>   http://www . itk . org/ItkSoftwareGuide . pdf
> > > >>
> > > >>    Section 8.13, pdf-page 318.
> > > >>
> > > >>The code is described there available at
> > > >>
> > > >
> > > >
> >
> Insight/Examples/Registration/DeformableRegistraton2.cxx
> > > >
> > > >>The output of the Demons filter is a deformation
> > > >>field
> > > >>(an image of vectors).  You can use this
> > deformation
> > > >>field
> > > >>as input to the itk::WarpImageFilter
> > > >>
> > > >
> > > >
> > >
> >
>
http://public . kitware . com/Insight/Doxygen/html/classitk_1_1WarpImageFilt
> > er
> > > .html
> > > >
> > > >>and resample the moving image with it.
> > > >>(this is illustrated in the example).
> > > >>
> > > >>If you used the atlas-image as the moving image
> > on
> > > >>the
> > > >>registration process, then the warped
> > moving-image
> > > >>will
> > > >>be equivalent to a segmentation of your data.
> > > >>(your data to-be-segmented should have been the
> > > >>fixed
> > > >>image during the registration process).
> > > >>
> > > >>You may find simpler to modify the example that
> > is
> > > >>described in the software guide.
> > > >>
> > > >>
> > > >>For visualizing your data using VTK, you simply
> > need
> > > >>to use the ImageToVTKImageFilter classes
> > available
> > > >>at
> > > >>
> > > >>     InsightApplications/Auxiliary/vtk
> > > >>
> > > >>and use the VTK class vtkImageViewer2.
> > > >>
> > > >
> > > >
> >
> http://www . vtk . org/doc/release/4 . 2/html/classvtkImageViewer2 . html
> > > >
> > > >>
> > > >>
> > > >>Please let us know if you find any problems,
> > > >>
> > > >>
> > > >>
> > > >>    Thanks
> > > >>
> > > >>
> > > >>
> > > >>      Luis
> > > >>
> >
> =3D=3D=3D message truncated =3D=3D=3D
>=20
>=20
> __________________________________
> Do you Yahoo!?
> Yahoo! SiteBuilder - Free, easy-to-use web site design software
> http://sitebuilder . yahoo . com