I am interested in taking medical data, such as MRI, CAT, etc, and
putting them together in povray to give a realistic view of the data in
3D. This can be done holographically, by multiply recording the slices
(e.g. http://www.voxel.com), but I want to see how good it would be in
povray.

There is a lot of work in detecting surfaces from the data, but all I
want to do is to stack them up and look at the whole thing, using stereo
or animation.

Questions:

1. Are there sample data sets? I have found one or two, but they are in
formats which I can't read. I would prefer tif, for example.

2. Has anyone done this already in povray? I don't want to reinvent the
wheel.

3. The fun bit is choosing the lighting for each slice, in order that
the slices add up to a good 3D image. Any hints on this?

-- 
Kaveh

Post a reply to this message

Kaveh wrote:

> I am interested in taking medical data, such as MRI, CAT, etc, and
> putting them together in povray to give a realistic view of the data in
> 3D. This can be done holographically, by multiply recording the slices
> (e.g. http://www.voxel.com), but I want to see how good it would be in
> povray.
> 
> There is a lot of work in detecting surfaces from the data,


http://astronomy.swin.edu.au/~pbourke/modelling/polygonise/

Post a reply to this message

Le Forgeron <jgr### [at] freefr> wrote:

> Kaveh wrote:
> 
> > I am interested in taking medical data, such as MRI, CAT, etc, and
> > putting them together in povray to give a realistic view of the data in
> > 3D. This can be done holographically, by multiply recording the slices
> > (e.g. http://www.voxel.com), but I want to see how good it would be in
> > povray.
> > 
> > There is a lot of work in detecting surfaces from the data,
> 
> 
> http://astronomy.swin.edu.au/~pbourke/modelling/polygonise/

Thanks for this link which is actually interesting and which I am
reading now. 

But actually I don't want to do any such computation, but just put the
planes up together and look at it as a whole, and let the eye/brain
decipher the data. 

-- 
Kaveh

Post a reply to this message

in news:1fk1p68.1yo4nh4w9gu0wN%kaveh@delete_this.focalimage.com Kaveh 
wrote:

> I am interested in taking medical data, such as MRI, CAT, etc, and
> putting them together in povray to give a realistic view of the data in
> 3D. 

http://news.povray.org/povray.binaries.images/13012/

has some of the information you're looking for. The MRI set is rendered 
directly as an isosurface (white). emitting media (pink & blue) is also 
used used.

The datasets mentioned are available here:
http://gd.tuwien.ac.at/visual/public.kitware.com/vtk/vtk2.4/data/


Ingo

Post a reply to this message

It sounds like what you want to do is map images with
transparency to planes. Should be easy enough to do.
Then you render stereoscopic pairs of the stacked planes
to view. Nice idea. You would have to be very precise
in registration of the images on the planes, and no
doubt you would have to play around with the levels
of transparency to get it to work.

Then again to really use the power of a ray tracer you
would want to transform the data into a true three
dimensional model. That would be a real challenge and
has been accomplished by Voxel and other companies
that charge huge sums of money for the process.

 HB


"Kaveh" <kav### [at] delete_thisfocalimagecom> wrote in message
news:1fk1v9v.1kv3c4e1dl2lc8N%kaveh@delete_this.focalimage.com...
> Le Forgeron <jgr### [at] freefr> wrote:
>
> > Kaveh wrote:
> >
> > > I am interested in taking medical data, such as MRI, CAT, etc, and
> > > putting them together in povray to give a realistic view of the data
in
> > > 3D. This can be done holographically, by multiply recording the slices
> > > (e.g. http://www.voxel.com), but I want to see how good it would be in
> > > povray.
> > >
> > > There is a lot of work in detecting surfaces from the data,
> >
> >
> > http://astronomy.swin.edu.au/~pbourke/modelling/polygonise/
>
> Thanks for this link which is actually interesting and which I am
> reading now.
>
> But actually I don't want to do any such computation, but just put the
> planes up together and look at it as a whole, and let the eye/brain
> decipher the data.
>
> --
> Kaveh

Post a reply to this message

ingo <ing### [at] tagpovrayorg> wrote:

> in news:1fk1p68.1yo4nh4w9gu0wN%kaveh@delete_this.focalimage.com Kaveh
> wrote:
> 
> > I am interested in taking medical data, such as MRI, CAT, etc, and
> > putting them together in povray to give a realistic view of the data in
> > 3D. 
> 
> http://news.povray.org/povray.binaries.images/13012/
> 
> has some of the information you're looking for. The MRI set is rendered
> directly as an isosurface (white). emitting media (pink & blue) is also
> used used.
> 
> The datasets mentioned are available here:
> http://gd.tuwien.ac.at/visual/public.kitware.com/vtk/vtk2.4/data/

Looks like *exactly* what I am looking for. Thanks Ingo. :-)

-- 
Kaveh

Post a reply to this message

This isn't quite impossible using only POV...

If I'm not mistaken, POV has some vector-function
which returns the rgb of a given pigment. Using
image-maps (these "slices"), you could somehow
sample the maps and get the "outer hull". Since
next layers should either line up when there
are multiple hulls present, the sampled points could
be connected using triangles, and otherwise there should be
some method to find if two hulls merge in a next layer...

But its a quite difficult process I believe. Perhaps making
Heightfields from image-maps, placing them, then tracing
samples could be another approach (though probably not
very accurate, but perhaps sufficient?)...

Ah, just mumbling about, don't mind me...


--
Tim Nikias
Homepage: http://www.digitaltwilight.de/no_lights/index.html
Email: Tim### [at] gmxde

> It sounds like what you want to do is map images with
> transparency to planes. Should be easy enough to do.
> Then you render stereoscopic pairs of the stacked planes
> to view. Nice idea. You would have to be very precise
> in registration of the images on the planes, and no
> doubt you would have to play around with the levels
> of transparency to get it to work.
>
> Then again to really use the power of a ray tracer you
> would want to transform the data into a true three
> dimensional model. That would be a real challenge and
> has been accomplished by Voxel and other companies
> that charge huge sums of money for the process.
>
>  HB
>
>

Post a reply to this message

Harold Baize <bai### [at] itsaucsfedu> wrote:

> It sounds like what you want to do is map images with
> transparency to planes. Should be easy enough to do.
> Then you render stereoscopic pairs of the stacked planes
> to view. Nice idea. You would have to be very precise
> in registration of the images on the planes, and no
> doubt you would have to play around with the levels
> of transparency to get it to work.

Well, the precision would be OK, as the data is all digital, and the
subject is fixed throughout. What I look forward to is learning and
using all the fine controls in pov to get precisely the effect needed.
(I am still a newbie in pov.) Ingo's frog looks beautiful. 

> 
> Then again to really use the power of a ray tracer you
> would want to transform the data into a true three
> dimensional model. That would be a real challenge and
> has been accomplished by Voxel and other companies
> that charge huge sums of money for the process.

Well, I know the Voxel guys *very* well, and the technical guys are good
friends of mine. I am now excited about getting POV to make a useful
image, using *no* money!

-- 
Kaveh

Post a reply to this message

Tim Nikias <tim### [at] gmxde> wrote:

> This isn't quite impossible using only POV...
> 
> If I'm not mistaken, POV has some vector-function
> which returns the rgb of a given pigment. Using
> image-maps (these "slices"), you could somehow
> sample the maps and get the "outer hull". Since
> next layers should either line up when there
> are multiple hulls present, the sampled points could
> be connected using triangles, and otherwise there should be
> some method to find if two hulls merge in a next layer...
> 
> But its a quite difficult process I believe. Perhaps making
> Heightfields from image-maps, placing them, then tracing
> samples could be another approach (though probably not
> very accurate, but perhaps sufficient?)...

I think you are thinking along the lines of what I *don't* want to do!
It's actually very simple. Think of, say, 10 slices of an image, each
one fully transparent, but emitting light. (Sorry, newbie to pov so
might not use the right terms.)

You will be looking at and through all the slices, and your eye/brain
will make these into a 3D image. 

You could do this simply by printing each slice onto a glass sheet.
Where there is no image, the glass is transparent. Where there is an
image, the image is, white, say. When you look at the whole set, you
will see a 3D image with full parallax as you move your head.

Except you can do it better in povray as you don't have the physical
constraints of one image hiding another. 

Hope you see what I am saying.

-- 
Kaveh

Post a reply to this message

On Mon, 14 Oct 2002 13:33:56 +0100, kav### [at] delete_thisfocalimagecom (Kaveh)
wrote:
> 2. Has anyone done this already in povray? I don't want to reinvent the
> wheel.

Is DF3 format sufficient ?
If you want operate on slices with intersections it could be simple to write
functions based on 2D image_patterns interpolated in 3rd dimension and applied
as pattern to media or as function to isosurface.

ABX

Post a reply to this message