Hi there,
Does anyone have advice on how to best use Shapeworks for vessels? I’ve been trying to create a good SSM of branch pulmonary arteries (and aorta), but so far I have not been successful even with smoothing (using Meshmixer), remeshing and trimming the branches to keep the anatomy as simple as possible. Mean template is still fairly coarse and the shape variations are jagged and non-physiologic. Many thanks in advance.
Hi,
Can you share the optimization parameters that you’re using? A screenshot of the optimization tab would be good.
If you watch the optimization, are the particles spreading well at each stage, before splitting? If not, I would recommend increasing the number of iterations per split.
Another thing to try would be to turn on “normals” for the optimization.
Thanks for responding! Here are my parameters - I increased number of iterations per split with still similar results. They seem to spread evenly, although in some situations they cluster together before the last iteration (I excluded those from this most recent run). You can see the type of PAs I am using too. I have previously tried using particles from initial landmarks, but didn’t see much obvious improvement.
It may be difficult due to the way that the branches go in different directions across the samples.
Is there a chance you could share this data with us so that we could experiment to see if we can help?
Sure thing - what is the best way of sending?
Usually a google drive, box, dropbox link. You can send to my email amorris@sci.utah.edu
I’ve experimented with this data a bit and recommend using “geodesics” as well as “procrustes”. With a low number of particles (64), things look pretty good:
Things start to get a little more challenging at higher particle counts, but I don’t know the level of detail that you need. If you are interested in modeling the branching structure, then a low particle count may be sufficient.
Here is 128 particles:
Oh interesting, thanks! I’ll play around with the Procrustes and geodesics then.
So far it’s looking much better - having constraints onto the outlets have also helped the SSM significantly.
I am having a new issue regarding using multiple domains in this context- in this case, I’d like to also do SSM of the respective right ventricles and pulmonary arteries together. It works, however I having issues with finding the appropriate alignment - when I try to align them with interative closest point together, it crashes (see screenshot below). I can align them with ICP within each shape cohort by loading a separate studios file, performing ICP, then copying and pasting the alignment column onto the main project file. However, there is no global alignment for ICP so I am left with center which doesn’t generate a good result. Any advice? I have shared with you a demo folder of 10 cases as well. Thanks!
This seems to be a bug that’s been fixed since the 6.5 release. Can you try the latest development release?
Happy to say that the multi-domain ICP is not crashing; however I’m still struggling with getting decent global alignment . I used ICP to align to the initial template (see last model), but there still to be some residual translational aspects that are not captured, and indeed the first shape mode still has that component when using global alignment (as opposed to the other alignment methods). Any advice on this, or should I just stick with using the aorta/pulmonary arteries as the reference alignment?
This is challenging when the structures have varying dimensions like this (some longer, shorter). I might suggest placing a few landmarks and using the landmark based alignment.
