Meshing problem with 'Aortofemoral Normal - 2' example

[Rok Oblak]

Hi SimVascular team,

I'm trying to create a surface mesh from the provided examples
http://simvascular.github.io/clinicalCase1.html
and
http://simvascular.github.io/clinicalCase2.html

The first one works fine - I import the paths and segmentations, create a model from the lofts, and then mesh the model to get the mesh-complete directory ready for simulation.

However, the second example is troublesome. I can create the model from the lofts and it looks fine in the editor, however when I try to mesh it it always fails, giving an error of duplicated vertices or similar. I usually get this error when the edge size is too large, but in this case even an edge size 3x smaller than the 'estimated' value (should be more than small enough, and it runs for 45 minutes on a modern CPU) gives the same error.


Is there any adaptation that I need to make on the model before it can be meshed properly?
I tried blending faces but SV crashes and I've seen in another thread this will be fixed in a future release, is there anything else I can try?

Thanks and best regards,
Rok

[Adam Updegrove]

Hi Rok,

No, there shouldn't be any additional steps needed. Sounds like there is something wrong in the model created. I typically do a surface remeshing prior to creating a full volumetric mesh to make sure that I'm giving the mesher a quality input surface.

If you don't mind sharing the project, I could take a look for you.

SimVascular Development Team

[Rok Oblak]

Hi Adam,

thanks for the help.
The following are two projects:
The "example_aorta" is from the "Aortofemoral Normal - 2" example (http://simvascular.github.io/clinicalCase2.html):

https://www.dropbox.com/s/4jdudar7b3yfj ... a.zip?dl=0

The problem with this one is that it won't create a mesh from a model (but it will create the model).

The "example3" is from the "Healthy Pulmonary" example (http://simvascular.github.io/clinicalCase4.html):

https://www.dropbox.com/s/9vry4y8j15ji3 ... 3.zip?dl=0

The problem with this one is that it won't even create a model from the segmentations.

Maybe the segmentations need to be adjusted a bit, but I don't know how and where.

Thanks and best regards,
Rok

[Adam Updegrove]

Hi Rok,

Thanks for the post.

For the aortafemoral example, I'm attaching an image which is the cause of the meshing error. The segmentations for this model were created in a previous framework. As you can see, the segmentations in the Aorta seem to be bunched together and out of order at the end of the vessel. This causes an overlapping issue, which would render the model unmeshable. We need to update and correct the segmentation groups for this model, and we will do so ASAP. Thanks for reporting!

For the pulmonary example, it appears you have found a bug! We recently updated our lofting algorithms, and although the vessels are being lofted correctly, there is an in between step that is going wrong for some vessels. Take for example LPA_07, in the segmentation panel, you can do the lofting preview and the vessel lofts fine. However, if you go to model creation and attempt to create a model with only the group LPA_07 segmentations, you get an empty model. This obviously causes issues when creating the full model. This bug will be fixed and corrected in the next post of SimVascular. For now, either use the old lofting methods, set under "Set Uniform Lofting Parameters" and change "nurbs" to "spline", or create a model with a subset of the branches. I found that LPA_07, LPA_19, LPA_35, and LPA_40 all have this issue, but there could potentially be more.

Thanks again for reporting, these issues will be fixed ASAP!

SimVascular Development Team

[Rok Oblak]

Hi,

thanks for the input.
Just as a sidenote, I found one other problem with the "Aortofemoral Normal -1" eample (http://simvascular.github.io/clinicalCase1.html): after the lofted model is done, there is a hole at a junction, see attached screenshots below.

Screen Shot 2017-07-20 at 5.20.33 PM.png (127.07 KiB) Viewed 965 times

Screen Shot 2017-07-20 at 5.20.20 PM.png (122.11 KiB) Viewed 965 times

Not sure this is a lofting problem since it seems more like a problem with the segmentation (either it should have been "narrower" at the end or further down the path), but still thought I'd point it out.


By the way, do you know about that tetgen flag I asked in a separate thread (viewtopicPhpbb.php?f=188&t=6853&p=0&start=10&view=)?

Hi,

one more question: what is the .tgs flag to use “Radius-Based Meshing”? Also, in one of the examples I saw the option "UseMMG", which seems to speed up the meshing, what exactly is this, is this the same as the "Use fast meshing" option in the GUI?

Thanks,
Rok

Thanks and best regards,
Rok

[Adam Updegrove]

Hi Rok,

Again, thank you making us aware! This is again an issue with the segmentations. Its not actually a hole, but just the end cap of the internal iliac branch. We will fix these segmentations as well.

If this project has the image data in it, you should be able to edit the segmentations yourself if you would like until we have a chance to fix them.

I'll respond to that in the respective post.

Thanks!

SimVascular Development Team

[Rok Oblak]

Hi,

As you suggested, using "nurbs" instead of "spline" for the uniform lofting parameter in the pulmonary example, I was able to get a model, but, just like the aortofemoral example, the meshing fails.

I tried to check the generated aortofemoral example model and besides the cap problem - the one you got on the screenshot - I didn't see any other meshing problems. Because I fixed that cap (I cut the mesh further down to get rid of the problematic wal faces) but the meshing still fails.

Do you maybe know if this will be fixed in a further SimVascular release (and when do you plan on releasing it)?

Thanks,
Rok

[Adam Updegrove]

Hi Rok,

Are you sure there were no extra cap faces? If you look of the names of the faces, its fairly obvious if there are additional caps because there will be an _2 at the end of the face name. You should have at least one, which would correspond to the main vessel. However, if there are multiple cap faces with _2 at the end, you should start highlighting to find where they exist. In future versions, we will indicate more clearly when this is occurring.

If you want, you can send me the project and I will take a look.

Thanks!

SimVascular Development Team

[Rok Oblak]

Hi Adam,

after I cut the problematic cap (and generated a new cap) I checked every other cap and everything seems OK, but I might have missed something.

The following error is thrown by tetgen:

PLC Error: A segment and a facet intersect at point (0,109787,0,0707731,0,527817).
Segment: [27621,65235] #-1
Facet: [27620,27610,27523] #1

Below is a link to a project where a model is imported, but the meshing fails:
https://www.dropbox.com/s/g3gwqriuxrpgm ... 1.zip?dl=0

Thanks,
Rok

[Adam Updegrove]

Hi Rok,

I usually do a global surface remesh in the modeling module to make sure my surface looks good before I take to the volumetric meshing module.

In modeling, this is under "Global" operations, "Remesh". I used the default estimated mesh size, and remeshed.

I then took to the volumetric meshing module, and ran the volumetric mesh with the estimated mesh size. This worked fine.

Here is the project if you're curious:

https://berkeley.box.com/s/p1lojjar7jkc ... t8eh570lk9

Thanks!

SimVascular Development Team