 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Here's a sample render of a better-than-average MRI (resolution-wise) scan
of my head. Lot's of artefacts visible, all due to the dataset rather than
POV-Ray.
I can make the .df3 (and the original DICOM dataset as well) available if
there's interest. A word of warning though -- it's a bit difficult, though
not at all impossible, to obtain better renderings than this from the MRI
data. It requires a lot of pre-processing (noise filtering, segmentation,
....) and the (free) tools available for this are a world of their own.
On a related note, has anyone heard of <a
href="http://www.kumpu.org/jto/phd.html">seed filling in view lattice</a>?
I wonder how it would fit within the POV-Ray media framework?
--
jussi.kantola
Post a reply to this message
Attachments:
Download 'mri9.jpg' (135 KB)
Preview of image 'mri9.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Here's a sample render of a better-than-average MRI
> (resolution-wise) scan
> of my head.
Very interesting.
I've had regular CT, MRI and PET scans. How large are the original MRI
files? I've asked for copies ( to play with in Povray of course :) but
the request always seems to get lost in red-tape. I was told
resolution is about 1mm, and was guessing I could convert and clean
the data to something useful.
Alf
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Alf Peake" <alf### [at] peake42 freeservecouk> wrote:
>
> I've had regular CT, MRI and PET scans. How large are the original MRI
> files? I've asked for copies ( to play with in Povray of course :) but
> the request always seems to get lost in red-tape. I was told
> resolution is about 1mm, and was guessing I could convert and clean
> the data to something useful.
It varies with the imaging protocol used, but the one I posted was from a
16bit 512x512x232 voxel dataset, which translates to voxel dimensions of
<0.39,0.39,0.9> (mm) for the field of view that was used in the scan. That
(uncompressed) image series takes about 180MB.
There are many free medical visualization/processing tools available on the
internet (e.g. AFNI, Dicom Toolkit, MRIcro ... ) you could use for
conversion and processing. There could be a bit of a learning curve in
using them, as medical images use special formats (mainly DICOM) and
terminology. The imaging methods are a whole different issue, and one has
to at least take a look at a radiology tutorial to understand what to
expect to see in different images.
Now, cleaning up the data isn't too difficult, and one could in principle do
it for the separate grayscale slices with a common image editor and then
convert the slices to a .df3. More advanced processing, like segmenting
out different subvolumes (gray brain matter, veins) is a lot harder, even
with tools specifically designed for the purpose. If you see a gorgeous 3D
rendering of a MRI dataset showing all the different organs with clean
color coding, you can be quite certain someone's spend a good few hours
processing the data to a workable condition!
To get your own images, you simply have to find the nerd of the place you're
being imaged at :) As it is easier said than done, I'll post at least the
..df3 and possibly a set of grayscale images next week when I get back to
faster connections.
--
jussi.kantola
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"jute" <nomail@nomail> wrote in message
news:web.415b2bba458f451d378d0fe0@news.povray.org...
> To get your own images, you simply have to find the nerd of the
> place you're
> being imaged at :)
Thanks for that info. I even took a blank CD with me the last time but
their boxes had no write facility :/ Mind you - I'm talking about UK
health service ;)
> As it is easier said than done, I'll post at least the
> ..df3 and possibly a set of grayscale images next week
I'm using landline so I tend to skip anything larger than a coupla
megs.
Alf
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"jute" <nomail@nomail> wrote in message
news:web.415aa11968f0cebfee6573000@news.povray.org...
>
> Here's a sample render of a better-than-average MRI (resolution-wise) scan
> of my head. Lot's of artefacts visible, all due to the dataset rather
> than
> POV-Ray.
>
> I can make the .df3 (and the original DICOM dataset as well) available if
> there's interest. A word of warning though -- it's a bit difficult,
> though
> not at all impossible, to obtain better renderings than this from the MRI
> data. It requires a lot of pre-processing (noise filtering, segmentation,
> ....) and the (free) tools available for this are a world of their own.
>
I did some renders from CT and MRI scans some time ago (using a df3 file
directly and converting it to a
thresholded mesh) The MRI scan quality that I obtained with a simple
threshold was far better than you are
getting here and the CT scans were astonishingly detailed. That they use
big, complicated applications
in hospitals doesn't neccessarily mean that a simpler approach won't give
good results. I can make the df3 files /
meshes / mesh conversion code available if anyone is interested. I've
included a render of one of the sectioned meshes
that I made (A CT scan in this case)
Andy
Post a reply to this message
Attachments:
Download 'Sectioned CTHeadIso 100.jpg' (38 KB)
Preview of image 'Sectioned CTHeadIso 100.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
in news:415e7489@news.povray.org Andrew Le Couteur Bisson wrote:
> I can make the df3 files /
> meshes / mesh conversion code available if anyone is interested.
Please do so. I'd be interested to use that in combination with Gilles
cloud system. I'd like to see wther using a mesh instead of a box as
container object speeds up the rendering.
Ingo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"ingo" <ing### [at] tagpovrayorg> wrote in message
news:Xns9576802EBE36Dseed7@news.povray.org...
> in news:415e7489@news.povray.org Andrew Le Couteur Bisson wrote:
>
>> I can make the df3 files /
>> meshes / mesh conversion code available if anyone is interested.
>
> Please do so. I'd be interested to use that in combination with Gilles
> cloud system. I'd like to see wther using a mesh instead of a box as
> container object speeds up the rendering.
>
> Ingo
I have a 6.5M zip file with sources and df3 files for the Stanford volume
rendering examples (cthead and mrbrain)
where should I upload/email these (I presently have no web space)
Andy
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
in news:415fcd5a$1@news.povray.org Andrew Le Couteur Bisson wrote:
> I have a 6.5M zip file with sources and df3 files for the Stanford
> volume rendering examples (cthead and mrbrain)
>
That's a bit too big to mail or post. Could you just post a zip of your
conversion code to povray.binaries.utilities ?
Thanks,
Ingo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"ingo" <ing### [at] tagpovrayorg> wrote in message
news:Xns95778CF7E2BB9seed7@news.povray.org...
> in news:415fcd5a$1@news.povray.org Andrew Le Couteur Bisson wrote:
>
>> I have a 6.5M zip file with sources and df3 files for the Stanford
>> volume rendering examples (cthead and mrbrain)
>>
>
> That's a bit too big to mail or post. Could you just post a zip of your
> conversion code to povray.binaries.utilities ?
>
> Thanks,
>
> Ingo
Done!
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
in news:41604e1a$1@news.povray.org Andrew Le Couteur Bisson wrote:
> Done!
>
Thanks
Ingo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
