[vtkusers] vtkImageData & tensors & filters

[David Gobbi]

Hi Alexandre, Christina:

I don't think that anything apart from perhaps some new imaging
filters is required to support tensor images.  Most of the
existing filters will already accept any number of components
per pixel, so you just need to use images with 6 (or is it 9?)
components as needed to describe the tensor.

Writing new filters will probably be both easier & cleaner
than trying to extend existing filters.  For example, how would
vtkImageMathematics know whether it was dealing with a tensor image
rather than just a string of 6 components?  A separate e.g.
vtkImageTensorMathematics filter would just assume that the data
being fed in was composed of tensors.

No changes to vtkImageSource should be necessary, as far as I can
see, as long as >4 components are already properly supported.

Could you make a list of operations you'll need to be supported for
tensors?  I'm not one of the 'vtk folks' but I can probably offer
some advice re: implementation.

 - David

--
  David Gobbi, MSc                       dgobbi at irus.rri.ca
  Advanced Imaging Research Group
  Robarts Research Institute, University of Western Ontario

On Thu, 28 Jun 2001, Chr. Rossmanith wrote:

> Hi,
>
> I think it would be very useful to have such operators for tensor
> valued image volumes, e.g. in the context of processing MR diffusion
> tensor image data.
>
> Christina Rossmanith
> --
> University Hospital Mannheim,
> University Heidelberg
>
>
> Alexandre Guimond wrote:
>
> > Hi.
> >
> > I'm dealing with tensors that are organized on what the vtk user's
> > guide calls "structured points", ie a rectangular, evenly spaced
> > grid.  Hence, I'm placing them in a vtkImageData.
> >
> > A few operations that I need to do on tensors are generalizations to
> > tensors of filters that exist in the imaging toolbox for scalars.
> > Examples of this are mathematical operations or shrinking.
> >
> > What I'm wondering now is if I should write new filters for tensors or
> > extend the vtk ones.  It's obviously not worth it for me to extend the
> > vtk ones if vtk is not willing to use the modifications.  Hence the
> > question: would the vtk folks be interested in code that generalizes
> > of a few imaging classes to cope with tensors, or is the goal to only
> > deal with scalar data in image filters?  That also implies a few
> > modifications to vtkImageSource, etc. for memory allocations
> > purposes, etc.
> > --
> > Alexandre Guimond, Ph.D. guimond at bwh.harvard.edu Phone:(617) 278-0800
> > http://spl.bwh.harvard.edu:8000/~guimond/          Fax:(617) 264-5154
> > Brigham and Women's Hospital, Harvard Medical School
> > CNI, RF396, 221 Longwood Av, Boston, MA 02115
>
>
> _______________________________________________
> This is the private VTK discussion list.
> Please keep messages on-topic. Check the FAQ at: <http://public.kitware.com/cgi-bin/vtkfaq>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/vtkusers
>