#MRI, CAT to DXF or 3DS?
16 messages in this thread
Hi everybody-
I need suggestions about how to convert three thousand 8 bit and 16 bit
greyscale 2D images that are 1 mm. thick CAT and MRI scan slices of the human
body into a dimensionally accurate 3D wiremesh file that I can animate in 3DS
or navigate in CDK.
This is new project for me, and I'm starting from scratch, so that ANY
information you might have about other people or institutions who have done it,
availability of UNIX based conversion programs, IPAS's (I wish), or scripting
possibilities would be greatly appreciated.
Even just a suggestion of where to begin….:^).
Thanks,
Carlton
Have you heard of the "Human Body" project? A corpse was frozen in a composite
and then sliced into thin segments and scanned in. From there, the data was
organized into a volumetric representation for study purposes.
I think there might be contact for these people on last year's Siggraph video
review. I'm sure they could offer some suggestions. I'd look it up myself, by
my tapes are loaned out right now.
— James — Choreo Motion —
James-
"heard of the Human Body Project…"
Sure have, even played with it a little on the www. I haven't asked them
though. A good idea.
There are obviously a lot of volume renderers around on the Unix boxes that the
doctors use. And I'm not sure that the shaded images I see are shaded
wireframes. My problem is how to get the 3D data into a form I can use with
Autodesk products.
I haven't started a thorough inquiry yet, still too many projects pending, but
hey, isn't the Ammedia forum where all the knowledgeable folks hang out??
:*)
Carlton
There is really no way to take the raw data files from MRI or CT and convert
them into a 3ds or dxf format. The main reason is that each company has its
own proprietary format and therefore you cant access it. There is some
standardization beginning but it is not in the near future. If you want to get
some 3d data from the actual scans there is a way to do it for the pc, you
cannot, however, take that data into 3d studio. There is a program called
spyglass which will allow you to take consecutive slices of the scan(ie- a mri
of the head in the axial plane will take about 14-16 3-5mm slices)convert the
actual slices into gif files, whichever way you want to, direct screen capture,
scanning, video etc. You then have to reposition each gif file so that if all
images are placed on the same plane the reference points stay exactly in the
same spot. These gif files are then brought into spy glass and depending on
the pixel values you can then render, raytrace, the areas which you want to
discern. It is not done in real time but it works out pretty well. You can
also animate it and apply false colors. We do some of this for our projects and
are currently working with one of the imaging vendors to allow us to directly
register the image files so we can import them directly into spy glass. Very,
very tedious. Hope this helps. Sam Coor Medical Media
Sam – Thanks for the info. Do you by any chance know how to get into touch
with the people that make the conversion program "Spyglass"?
Also I was wondering….I can get 8 bit greyscale PCX files converted from the
MRI's…I could imagine drawing contours around each distinct grey area and
making a shape and then lofting them in the lofter (Ugh! Talk about tedious.
:*)…Do you know of any raster to vector conversion program that might animate
this part of the process?
BTW, I work at Virginia Commonwealth University, and the projects that might
come out of this data conversion would all be for research and education in
conjunction with our radiology department. They have some very interesting
imaging equipment…but we've got 3DS!
Carlton
I will get the info on spyglass. I do not know of any raster to vector
software that is that good. In terms of tracing contours, whew good luck-been
there, tried that and do not want to revisit. In terms of directly converting
mri to 3ds there is some major research going on in this field. I just came
back, last week, from a conference on Medical Informatics at Stanford and they
spent about a day going over image processing. One of the major problems is
converting mri ct to actual mesh files they went over a number of contour
finding algorithims and there strengths and weaknesses and so far none fit the
bill. Interestingly they are beginning to use neural nets in there contour
algorithims and this looks like where it will head. One thing to note which
was really amazing is that they took CT data and then volume rendered it and
flew through the areas. An example they did spiral CT scanning through the
sinus and then did a fly through just as if you were using an endoscope. So I
guess the big question is by the time the right algorithms are determined to
convert mri to a 3d mesh the 3d reconstructive fly throughs will probably be
standard of care. Sam
>>I do not know of any raster to vector software that is that good
What I've heard second hand is that Adobe's Streamline is the best out there on
the PC; FWIW.
John Stetzer
JWS
Hi John,
I've played a litte with Adobe streamline, and it does seem to work well.
However the images the MRI scanner software makes are rather detailed 16 bit
greyscale, and its hard to get the program to work with such levels of detail.
We also need the process to be automated.
Carlton
Sam,
Could you also send me the info on Spy Glass? I'm very interested in where I
can obtain it and how much it costs.
Thanks,
Mark
Check your e-mail Sam
Thanks, Sam
Mark
>>Spyglass…
Carlton, I have two packages here, Slicer by Spyglass, and Voxelbox (volume
renderer) by Jaguar Software. It's been a while so I can't remember the
details, but I remember some kind of deal in the works between the two
companies with more 3D volume software comming. At any rate, the boxes say
Spyglass, 217-355-6000; Jaguar, 617-729-3659. (just in case you didn't get
numbers yet)
Sam, thanks again for all the information and especially for the Spyglass
telephone numbers. I spoke last Friday with a doctor who attended the same
Stanford medical informatics meeting and he too was very impressed with the
volume fly through.
Our university uses scanners from Seimens and Toshiba (and perhaps others)
and I guess one of my next steps is to contact their software people to see
how far along any conversion programs might be. My understanding was that
the stuff shown at Stanford was funded by the NIH (National Institute of
Health), a major grant project, and I suppose it is all being published.
Do you by any chance remember the name of the doctor or group that did the
fly through?
Thanks again,
Carlton
I will check this weekend all of my course materials are at home. In terms of
being published everything the University does is basically publically funded
and therefore can be found in various journals and newsletters. Anyway I will
post the info Sam
Sam-
Just did a download of your CIS number from the libraries…So you're the guy
with the vertebre! And the brain too! I downloaded your images a while back
just to see what they looked like. They look very cool.
How is the skeleton projet going? Do you have a list of programs your company
has made that are for sale that you could send me e-mail. I'd like to show
them to my doctor friends.
Carlton Newton
The skeleton is finished and has been for some time. We have decided for now
to hold off on outside distribution. This should change shortly. It mainly has
to do with the parties involved that completed it and when we sold the
vertebral column some of the people were not directly in the medical field and
the data was too dense. In terms of programs most have been for internal use.
We do have one commercial program being distributed by the University of
Washington-"Neuroanatomy Review-The Spinal Cord". However after working with
an academic setting I found that it was not the best way to get products
distributed, updated and marketed appropriately. That program is kind of a
dead issue and it is about 1 year old. We are currently involved in a very
ambitious CD-ROM on functional anatomy, sports medicine, and biomechanics of
one of the major joints. Of course using 3d studio but with some SGI input.
This should probably take a few more months(I hope!). In terms of the
neuroanatomy program if you e-mail me your address I will try to see if I can
send you a copy. Its disked based. Sam