Create a .stl file from CT Scan data?

Post your advice, tips, suggestions, etc...
Post Reply
Lateralg
Posts: 138
Joined: Mon Apr 14, 2014 8:21 am

Create a .stl file from CT Scan data?

Post by Lateralg » Sat Apr 19, 2014 1:04 am

I think I can get the data, which as I understand is a series of images of slices of a part of our body.

Does anyone have an idea on where to start on a mission to create an .stl file from CT Scan data?
Gary
Make Better Things
Make Things Better

Lateralg
Posts: 138
Joined: Mon Apr 14, 2014 8:21 am

Re: Create a .stl file from CT Scan data?

Post by Lateralg » Sat Apr 19, 2014 1:48 am

I found a starting point ... ImageJ.

Tons of info there, more than I can get thru tonite. Looks like it's possible, but I lack the computer background to understand everything I read there.
Gary
Make Better Things
Make Things Better

Toby
Posts: 330
Joined: Wed Apr 09, 2014 9:44 pm

Re: Create a .stl file from CT Scan data?

Post by Toby » Sat Apr 19, 2014 1:54 am

Here's a link about it:

http://www.wired.com/2013/04/3d-printed-skeletons/

Unfortunately it seems there may be some sort of paywall on the info, even though the claim is that all the software used is free.

My guess is this is done with voxels. Each tissue has a characteristic density, which maps to a particular shade in the scan (I think). Using that it's easy to construct a 3d bitmap (i.e. a voxel representation) for any given tissue type, and then there's software for converting voxel-based models into stl files.

cliffwarner
Posts: 16
Joined: Sun Aug 03, 2014 1:11 pm

Re: Create a .stl file from CT Scan data?

Post by cliffwarner » Mon Aug 04, 2014 3:30 am

To convert CT data to a STL file of the region of interest, typically an organ, requires a program that performs what is called segmentation. This requires a lot of work starting with image processing of the 3D data working with contrast and leveling. The objective is to produce contrast between what you want to image and the rest of the anatomy. One then separates that 3D region out with masks and spends hours cleaning up the data. AT that point one has a set of voxels, 3D pixels, for the set you want. This can then be meshed into a STL file or Nurbs surfaces, etc. There are a number of commercial codes that do this well. We use ScanIP by Simpleware at work for this. I work with this just enough that it is painful.

-Cliff

davidb
Posts: 9
Joined: Fri Jun 06, 2014 4:57 pm

Re: Create a .stl file from CT Scan data?

Post by davidb » Tue Aug 05, 2014 2:48 pm

In addition to the already mentioned ImageJ (which is useful for format conversion and filtering, but cannot, as far as I am aware, export isosurfaces in a usable format), I'd take a look at OpenDX and meshlab. OpenDX will let you create isosurfaces AND export them in some form of standard format (I've used vrml in the past). Meshlab will read, clean, and convert meshes to a variety of formats including stl. Both are free, open source, and available as Ubuntu/debian packages. There is definitely a windows installer for meshlab, but getting OpenDX running on windows is likely to be painful. OpenDX is also somewhat arcane and has a fairly steep learning curve.

Depending on programming skills and inclination, mayavi might also be an option (you can do a lot graphically, but I expect you might need to write a couple of lines of code to export the scene in a useful format).

davidb
Posts: 9
Joined: Fri Jun 06, 2014 4:57 pm

Re: Create a .stl file from CT Scan data?

Post by davidb » Tue Aug 05, 2014 8:21 pm

Actually, ImageJ will do everything you want ...

- download Fiji (a batteries included version of ImageJ)
- open your data (use file->import->bio-formats if you have a compressed dicom image or something else which isn't handled natively)
-select plugins->3d viewer
- from within 3d viewer, select edit->display as->surface
- then edit->attributes->adjust threshold - set to something reasonable (will require trail and error)
- when you're happy, file->export surface as

Post Reply