Variable radius stent

[lassoan]

Current situation

The SDFStent capsule-chain SDF uses a single shared radius, so deployed stents are always
uniform-radius "pills". Simulating flared (funnel/trumpet) stent ends - or other
variable-radius devices - is not possible. No tracking issue; developed for and already
exercised by the SlicerSimVascular extension
(https://github.com/lassoan/SlicerSimVascular/tree/flared-stent).

Release Notes

• capsule_sdf, smin_sdf_capsule_contact_sculpt and compute_sdf_contact_displacements
  now accept the stent radius as a per-vertex array (linearly interpolated along each
  capsule segment) in addition to a scalar, enabling arbitrary radius profiles along the
  centerline (flared ends, tapers, local narrowings).
• New cap_height_fraction parameter flattens capsule end caps into half ellipsoids;
  this lets concave profiles deploy without wide capsules' spherical caps filling in
  neighboring narrow regions, and keeps a flared end from inflating a ball beyond the stent.
• New profile helpers in svmorph.core: stent_radius_profile (control points) and
  flared_stent_radius_profile (smoothstep flare).
• New deploy_stent options: --flare-R, --flare-length, --flare-end,
  --cap-height-fraction.
• Not breaking: all defaults reproduce the previous behavior exactly (scalar radius,
  spherical caps); existing call sites are unchanged.

Documentation

All new/changed functions have NumPy-style docstrings; a flared-stent
deployment example is added to README with parameter explanations.

Testing

New tests/test_deformation.py covers scalar/array radius equivalence, per-segment radius
interpolation, spherical vs. flattened cap geometry, and preservation of a concave
(dumbbell) profile; tests/test_geometry.py extended for the profile helpers. Full suite
passes (35 passed).

Code of Conduct & Contributing Guidelines

• I agree to follow the Code of Conduct and Contributing Guidelines.

[jeffbli]

Hi @lassoan , I took a closer look at the behavior of this flared-stent extension, I have two questions:

1. It seems like the current CLI only supports flaring one end of the stent at a time, and the full capability of specifying a variable radius profile is not yet available in the CLI or the GUI? I think the user may usually want to model a flared stent is flared at both ends?
2. More importantly: I tried the current one-end flare functionality on an example geometry and rendered the resulting stent SDF to visualize it. I noticed that the current half ellipsoid based construction results in a wavey artifact throughout the stent SDF itself, which in turn results in the expanded surface also having a wavey surface?

(vanilla stent, flared stent with spherical end cap, flared stent with half ellipsoid end cap, final surface)

[lassoan]

It seems like the current CLI only supports flaring one end of the stent at a time, and the full capability of specifying a variable radius profile is not yet available in the CLI or the GUI? I think the user may usually want to model a flared stent is flared at both ends?

Good point, I agree that it could be useful to expose an option on the CLI to make both ends flared. The full radius array can only be specified via the library API (this is what the Slicer module uses).

I only added GUI to the Slicer module, I don't intend to modify the GUI in the svMorph package.

More importantly: I tried the current one-end flare functionality on an example geometry and rendered the resulting stent SDF to visualize it. I noticed that the current half ellipsoid based construction results in a wavey artifact throughout the stent SDF itself, which in turn results in the expanded surface also having a wavey surface?

Depending on the point density and cap height indeed the field can be wavy. I always provided a combination of these that avoid waves, but I can change the algorithm to avoid the waves even if the profile is specified with a sparse point set.

I'll push updates to this branch soon.