Dear all,
I am developing a biomechanical model to estimate prostate deformations during brachytherapy. I am working with finite elements on a tetrahedral mesh derived from segmented images. I am trying to figure out currently how from the deformed mesh (result of finite element analysis) i will go back to the volumetric image (deformation of the initial image).
Any insight of somebody who have adressed similar tasks or any references would be very helpful.
Thank you in advance,
Konstantinos.
Hello,
What you could try is (if your image are from ct scan or mri data) :
- Export a list of nodes coordinates data after simulation from the FEM software to a .txt file or .xls
- Convert them to the same image space as your segmented image, using interpixelspace , interslicespace found in the dicom
- Round the coordinate list, this will give you the index of every pixel representing the deformed mesh
- Change the pixel value of the original image using the previously found index to display the deformed shape on the image.
The rounding may lead to the creation of gaps, but this can be corrected if it's annoying.
Depending on your segmentation software you can directly import stl files and convert them to mask. So that you will just have to export your deformed mesh to a .stl file and import it back to your segmentation software.
Hope it helps.
Jean-Baptiste
Hi,
You can give a try to the following method:
http://www.mathworks.com/matlabcentral/fileexchange/27390-mesh-voxelisation
Note that, since the method is designed for triangular-polygon meshes, you will need to first convert your tetrahedral mesh to a triangular one.
Hope it helps.
When I have needed to do this I have used the ITK and VTK libraries. I always use the C++ version, but they have python and Matlab libraries as well. You need to extract the surface from the volumetric mesh using a vtk class (if the extensions are compatible) and then you can use itk::TriangleMeshToBinaryImageFilter to create your mask. If you only need to do it once, you can use a software, such as Paraview to do the job.
If you are going to try what Jean-Baptiste said, use imfill from Matlab to fill up the gaps or a close operation (imclose in matlab), as last resort you could even use a median filter but this will increase the size of the image.
Hope it helps
They are several library i.e vtk or or small free packages (voxelizer from Standford, binvox from Priceton,..). Just type "mesh to voxe" in your l search engine
Try this papers:
1- Extraction of human mandible bones from multi-slice computed tomographic data
2- New segmentation approach to extract human mandible bones based on actual computed tomography data
3- Computer assisted determination of mandibular cystic lesion volume from computed tomographic data
4- A new numerical model to analyze stress distribution of TMJ disc from 2-D MRI scans
Hello!
Once I had a mesh humam model and I wanted to convert it into voxels. I used 2 different tools. The first one was a software called BINVOX. You can find easily googling its name and I wrote an article (https://www.researchgate.net/publication/282815661_Voxelizao_de_um_fantoma_MESH_para_acoplamento_ao_EGSnrc) explaining how to do this (it is in Portugues, but the main part you need is the Binvox's window to understand). The second tool I used came from my research group. They developed a software because we all work with voxelized geometries. This is our main site www.dosimetrianumerica.org/ and you can contact my advisor (José Wilson Vieira) to get the DIP software (Digital Image Processing). It is for free and a user manual is going to be written soon.
Hope it will be usefull ;)
Conference Paper Voxelização de um fantoma MESH para acoplamento ao EGSnrc
You can have a look at the convex hull code on skimage.
http://scikit-image.org/docs/dev/auto_examples/plot_convex_hull.html
Quick and easy test can verify whether it can work for 3d points as well.
I know an answer to your question if you are using AMIRA software.
Amira has a tool calling surface to label field allowing to convert surface to the label field and than to fill the closed inner volume with labels using symply few additional steps in a segmentation tool.
you have to try with VTK (visualization toolkit)
Test with Regular Grids or ImageGrid clases
- Badges
- Science topic