[MFit]

Yes its working correctly!

[MFit]

I fixed it, with your help!

So indeed the config.h.in file needed to be renamed to config.h. That was problem 1 solved. Then I got the following error:

/usr/local/include/mpfr.h:326:42: note: expanded from macro 'MPFR_DEPRECATED'
# define MPFR_DEPRECATED __attribute__ ((deprecated))
                                         ^
In file included from /Users/mfit/SOFA_PlugIn/SkeletonPlugin/MeshSkeletonization.cpp:31:
In file included from /Users/mfit/SOFA_PlugIn/SkeletonPlugin/MeshSkeletonization.inl:31:
In file included from /Users/mfit/SOFA_PlugIn/SkeletonPlugin/MeshSkeletonization.h:52:
In file included from /Users/mfit/Downloads/CGAL-4.7/include/CGAL/extract_mean_curvature_flow_skeleton.h:30:
In file included from /Users/mfit/Downloads/CGAL-4.7/include/CGAL/Mean_curvature_flow_skeletonization.h:54:
In file included from /Users/mfit/Downloads/CGAL-4.7/include/CGAL/Delaunay_triangulation_3.h:37:
In file included from /Users/mfit/Downloads/CGAL-4.7/include/CGAL/Triangulation_3.h:47:
In file included from /Users/mfit/Downloads/CGAL-4.7/include/CGAL/spatial_sort.h:25:
/Users/mfit/Downloads/CGAL-4.7/include/CGAL/hilbert_sort.h:54:9: error: 
      no member named 'random_shuffle' in namespace 'std'; did you mean simply
      'random_shuffle'?
        std::random_shuffle(begin,end, rng);
        ^~~~~~~~~~~~~~~~~~~

Indicating to me that there was something wrong with the CGAL install. Instead of using the CGAL4.7 installed at /usr/locaal/Cellar/cgal4.7, it used the one in my download file. CGAL_DIR is specified in the cmake file. I removed all previous installations or and files of CGAL, and brew installed CGAL 5.0.2. This did the trick.

Should I inform the author of the Skeleton plugin about this?

Thanks again!

[MFit]

Hi @Guillaume and @Hugo,

Thanks for the feedback. I installed cgal4.7 with homebrew, seems to work fine.

I tried installing the skeleton plugin without the CGALPlugin earlier and just now, and there were no errors in configuration and generation. However, when I make the build (on top of a clean build) in the terminal I get the following error:

/Users/mfit/sofa/PlugIn/SofaSkeletonPlugin/SkeletonPlugin/initSkeletonPlugin.cpp:22:10: fatal error: 
      'SkeletonPlugin/config.h' file not found
#include <SkeletonPlugin/config.h>
         ^~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated.
make[2]: *** [external_directories/SofaSkeletonPlugin/SkeletonPlugin/CMakeFiles/SkeletonPlugin.dir/initSkeletonPlugin.cpp.o] Error 1
make[2]: *** Waiting for unfinished jobs....
/Users/mfit/sofa/PlugIn/SofaSkeletonPlugin/SkeletonPlugin/MeshSkeletonization.cpp:30:10: fatal error: 
      'SkeletonPlugin/config.h' file not found
#include <SkeletonPlugin/config.h>

The external location of the plugin folder of Skeleton plugin is recognized (I think) and the CMakeLists.txt is opened. The config.h file is not found, giving me the impression that I am missing a crucial step in the configuration fase in CMake.

Any idea’s on continuing? The Skeleton plugin looks very promising.

Thanks again,
MFit

[MFit]

Wow, I never expected it to be this fast! Gonna look into it right away. Thanks for the follow up on this question.

Kind regards,
MFit

[MFit]

Dear Hugo,

This sounds great! I am looking forward to working with this package. The results obtained as described in the paper are very promising. I think this might be a nice asset to SOFA as well. Vessel branch simulations sounds like a package wanted by more than a few.

Kind regards,

MFit

[MFit]

Hi Hugo,

So yeah, the role of the aorta is minimal, as it is just a fixed beam.

Thank you for the reference, I will look into it!
The main problem is the amount of bifurcations, the current model has 10 of them. Which is still less than hepatic veins/portal veins/hepatic arteries.

Getting a skeleton representation of my vessel is not gonna be a problem. The MRAs used are easily segmented and skeletonized in 3DSlicer. Creating beam structures inline with this centerline should be easy. I am gonna look into this as well. Thank you for the tip! Any more info is welcome of course. I am guessing the original author, Star War, is also interested in this subject.

Kind regards,
MFit

[MFit]

Hi Hugo,

Yes its a part of the arterial system. Additonal info; the aorta is a fixed object. Currently I’ve got that covered through a ROIBox and box constraint right on top of the complete aorta. I think it is probably better to make the renal and segmental arterys a seperate object and link the aorta with the seperate object through an attach constraint, but thats just a side note.

The question would be, what is the best way to create a FEM for a vascular tree? Is that through a beamFEM with manual positioning of the beams or my approach with a tethahedromFEM (which does not have the best mechanical properties but it is completely automatic)?

Kind regards,
MFit

[MFit]

Hi Star Wars and Hugo,

Ive got a follow up question. I also work with vessels, a renal artery tree. Anatomically, a aorta has at least 1 bifurcation for the kidney, which branches into segmental arterys which also bifurcate.
Currently, I use a tetrahedron FEM but, for obvious reasons, this allows too much movement in unwanted diretions. Would you recommend creating a beamFEM force field and manually make sure all the different bifurcation positions are alligned considering there are no built in tools to do this? I need the arterys inside the kidney to individually move with the kidney so simplification of the anatomy (e.g. 1 beam) is not an option.

[MFit]

The QR is passive, yet the kidney should be following the QR code square. In a further project, the position of the QR code will determine the position of the kidney.

Right now Im thinking of implementing it as a BilateralInteractionConstraint. The mass of the square will act as a force going against the kidney, which makes sense.

Any feedback is welcome.

[MFit]

Heres a picture of the scene for further explanation:

https://imgur.com/F9fLuvy

QR rigid square in red, kidney in gray.

Kind regards,

MatFit

[MFit]

Heres a picture of the scene for further explanation:

View post on imgur.com

Kidney in gray, QR rigid square in red.

Kind regards,

MatFit

[MFit]

Ive got a follow up question, @Hugo.

I got the mesh working, and now included a rigid 3D square (its the base for a future QR code). Ive used the GenerateRigidMass engine for this purpose. Now I want to “glue” the rigid square with the earlier mentioned kidney. The square needs to stay on the exact location in relationship to the kidney. The kidney should be able to move anywhere except for through the square. The kidney should be “hanging” on the rigid element.

Ive come up with a couple of ideas, but thus far Ive had no luck implementing them, probably for some arbitrary mistake:
1. Create a fixed plane constraint to the kidney at the exact location of the square OR something like a ProjectToPlaneConstraint and link it to the topology of the square?
2. Use submultisetmapping to map the square’s mechanics to the kidney mechanics.
3. Use AttachConstraint to attach the rigid mechanics to the kidney mechanics. (This makes most sense to me, as you Attach the rigid mechanics and use it as a constraint for the deformation of the kidney)
4. DeformableOnRigidFrameMapping, using a rigid frame build with the size and position of the square (this doesnt seem like the right way to go).

Have you got any tips or directions which way to go? I can share the current .scn if you like (with OBJs) for further explanation. Might be a beginners question, but slowly but surely Ill get the hang of it. 😉

Kind regards and thanks in advance,
MatFit

[MFit]

Hi Hugo,

Thanks for the response. This has helped out a lot. I will work on fine-tuning the mechanical properties of the model for now and see where that gets me. The training does sound interesting, I will definitely take a look.

Kind regards,

MatFit

[MFit]

Hi guys,

Could I get a heads up? Ive increased the possible download time of the STL/OBJ/MSH files by 5 days.

Thanks in advance,

MFit

[Author]

Posts