# SSM of Branched Vascular Patient-Specific Geometries \_Branch Distortion

**URL:** <https://shapeworks.discourse.group/t/ssm-of-branched-vascular-patient-specific-geometries-branch-distortion/400>\
**Category:** Uncategorized\
**Created:** [March 19, 2026, 7:55pm UTC](https://shapeworks.discourse.group/t/ssm-of-branched-vascular-patient-specific-geometries-branch-distortion/400 "2026-03-19T19:55:03Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Nafis](https://avatars.discourse-cdn.com/v4/letter/n/8c91f0/32.png) [@Nafis](https://shapeworks.discourse.group/u/Nafis)\
**Post date:** [March 19, 2026, 7:55pm UTC](https://shapeworks.discourse.group/t/ssm-of-branched-vascular-patient-specific-geometries-branch-distortion/400/1 "2026-03-19T19:55:03Z")

</div>

Hi,  
I am currently working on a project to build SSM for branched vascular geometries, specifically models such as the aorta or coronary arteries. My goal is to obtain reliable PCA modes, generate new geometries from the learned modes, and eventually extract physiologically meaningful shape features capturing geometric variability.

At the moment, I am testing the workflow with a small dataset of only 4–6 aorta models before scaling to a larger cohort. However, in the Analyze stage I am not obtaining anatomically realistic shape variations. The generated models often show branch vessels appearing clustered together and non-physiological roughness, sharp deformations and irregularities in the trunk as well.

I suspect the issue may be related to my preprocessing, grooming, alignment, or optimization parameter choices, but I am not fully sure which step is contributing most.  
I have already tried adjusting several grooming and optimization parameters.

I would greatly appreciate guidance on which grooming and optimize steps are most critical for branched vascular anatomies, and recommended parameter ranges for vascular SSM cases, and whether ShapeWorks has examples or best practices specifically for cardiovascular branched arteries.

I have shared some screenshots as well. I can share the models if someone wants to try as well.  
Thanks a lot for the guidance & support.

_- Nafis Arefin_

 ![PCA Std. Dev. -1.5](https://global.discourse-cdn.com/free1/uploads/shapeworks/original/1X/73676a1f441db474db42f219c929cde2259ef4aa.jpeg)

 ![PCA Std. Dev. 0](https://global.discourse-cdn.com/free1/uploads/shapeworks/original/1X/85c769f721b031b38899906034ec88f154febcfc.jpeg)

 ![PCA Std. Dev. +1.5](https://global.discourse-cdn.com/free1/uploads/shapeworks/original/1X/bd918822648735736cdf169e29958b562a5fa3f2.jpeg)

 ![optimization parameters](https://global.discourse-cdn.com/free1/uploads/shapeworks/original/1X/74afba797e906ce9297a661006a9099e459c801e.jpeg)

---

<div class="post-metadata">

**Author:** ![amorris](https://avatars.discourse-cdn.com/v4/letter/a/73ab20/32.png) [@amorris](https://shapeworks.discourse.group/u/amorris)\
**Post date:** [March 25, 2026, 4:35pm UTC](https://shapeworks.discourse.group/t/ssm-of-branched-vascular-patient-specific-geometries-branch-distortion/400/2 "2026-03-25T16:35:49Z")

</div>

Hi,

You may find these references useful:

> [@Creating An SSM for Aortic models](https://shapeworks.discourse.group/t/creating-an-ssm-for-aortic-models/303/4):
>
> So the main issue here seems to be the scale of the data. For example, I examined one mesh and found the x values ranged from about 0.98 to 1.5. Unfortunately, ShapeWorks has some issues with very small and very large scale data that we are working on resolving. Right now, the best range is 0-100 or so. If these are human aorta in mm range, then they should have already been in a good coordinate space. I scaled your meshes by 100 and played with the parameters a bit. It’s not the best mode…

> [@Advice on SSM of branched vessels?](https://shapeworks.discourse.group/t/advice-on-ssm-of-branched-vessels/255):
>
> 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.

I would recommend trying the development version of ShapeWorks:

> **[Releases · SCIInstitute/ShapeWorks](https://github.com/SCIInstitute/ShapeWorks/releases)**
>
> ShapeWorks. Contribute to SCIInstitute/ShapeWorks development by creating an account on GitHub.

In particular, enable “geodesic distance”, “normals”, “procrustes” and “sampling scale”.

 ![image](https://global.discourse-cdn.com/free1/uploads/shapeworks/original/1X/96fc55044ff6c4c64e2b4efba10266443cf0c656.png)