For my graduation project I am working with FEM analysis of bone. I have seen and read a lot of threads regarding this problem, but not yet found a proper solution. I have a stack of 2D images from a CT scan which I need to convert to a 3D model. Since this model is used for mechanical analysis, and not printing or visualization, I also need to obtain nodal density values for modulus calculation.
A similar question is asked here: https://www.researchgate.net/post/Converting_CT_scan_images_into_volume_mesh2
but it doesn't seem to present a conclusive solution.
I need a non-binary representation of each voxel, for representing material parameters.
I am running the FEM analysis in Abaqus
I have looked into Fiji and 3D slicer, but they don't seem to provide the tools I need.
I also have access to Matlab, Siemens Nx and Solidworks, if anyone should have any experience here.
This process has previously been performed at my institute using Mimics from Materialize, but that software is no longer available to me. And also, I'm going to do this job multiple times, so it wouldn't be feasible to have someone with a license to do it for me...
Is there anyone who have done this before and could share their experience?
Thanks!
Felix Hamza Lup thank you! I see you have used VTK. I am not familiar with this. Would you please explain to me how to get started with this toolkit?
For mechanical analyses you need voxel based solid models for finite element analysis. All others and stl based models will be fall down in FEA. You can check my bio, I am so busy nowadays if not I can help your models. If you want to work in collaboration with others, for writing article type production I can guide you to contact some researchers. Have a nice day.
There is no standard perfect solution and there are several more threads here on RG discussing this problem.
I have used the following workflow:
create an STL surface mesh with 3dslicer or Fiji
Read this into autodesk meshmixer
Remesh there with a regular surface mesh, export it
Mesh the result with graphite (which internally uses tetgen)
http://alice.loria.fr/index.php?option=com_content&view=article&id=22
But this is not a foolproof way and often fails, requiring you to fiddle around with the stl mesh (in meshmixer or meshlab or some other software).
I have my in-house code to generate voxel-based FE models from 2D images for some commercial packages (such as Abaqud amd LS-Dyna).
It’s also possible to consider the variable elastic modulus/density! However, I may not be able to offer any help by the end of 2020.
Sajjad Raeisi Thank you for offer, but I do unfortunately not have that time.
Martin Baeker I am starting to realize why commercial softwares are pricing their licenses so steep. I'm working on a similair workflow myself:
1. Create STL from Fiji (filling all cavities in order to make a simpler mesh)
2. Removing islands with Meshlab, but no further remeshing. If you have any comments on necessary mesh refinements or remeshing, please share!
3. Attempt to turn mesh to solid in FreeCAD as explained here: https://www.bantamtools.com/blog/convert-stl-to-step
Then I will try to import all original scan images to matlab and create a point cloud with intensities, which I can use as reference node densities for the Abaqus mesh.
- Badges
- Science topic