Statistics via Python API

1

Hello ShapeWorks!

Is it possible to use the Python API to compute a mean shape, median shape, and other statistics?

For example, something like the follow would be extremely useful:

model = shapeworks.load_model("myssm.xlsx")
mean_particles = sw.stats.mean(model)
median_particles = sw.stats.median(model)

(A related question: Is the Python API fully documented somewhere?)

thanks!

2

The python documentation is here:

There is an Analyze module available that accomplishes some of this:

import shapeworks as sw
project = sw.Project()
project.load("d2_pre_only.xlsx")
analyze = sw.Analyze(project)
mean_particles = analyze.get_mean_shape_points()

But it looks like there isn’t yet a method to return the median shape particles. It should be pretty easy to add if desired.

3

Thanks Alan! I’m not sure how I didn’t find the documentation, but now it is clear.

4

Hey, Can you please tell how to get the shape meshes from these particles using python?

5

Mesh reconstruction from particles in ShapeWorks is typically done using mesh warping. A reference template mesh is used to warp based on the particles.

For example:

from shapeworks import *
reference_mesh = Mesh("template.vtk")
particles = ParticleSystem("template_particles.particles"]).ShapeAsPointSet(0)

warper = MeshWarper()
warper.generateWarp(reference_mesh, particles)

target_particles = ParticleSystem("target_particles.particles"]).ShapeAsPointSet(0)
warped_mesh = warper.buildMesh(target_particles)
6

Thank you. But with the below code,
from shapeworks import *

shape_path = ‘/mean_geo/par_512.stl’
shapeMesh = sw.Mesh(shape_path)
shapeMesh_vtk = sw.sw2vtkMesh(shapeMesh)
mean_particles = '/particles/mean_particles.particles’

reference_mesh =shapeMesh_vtk # Mesh(shapeMesh_vtk)
particles = ParticleSystem([mean_particles]).ShapeAsPointSet(0)

warper = MeshWarper()
warper.generateWarp(reference_mesh, particles)
it is showing me error as ’ generateWarp(): incompatible function arguments. The following argument types are supported:
1. (self: shapeworks_py.MeshWarper, reference_mesh: shapeworks_py.Mesh, reference_particles: numpy.ndarray[numpy.float64[m, n]]) → bool
2. (self: shapeworks_py.MeshWarper, reference_mesh: shapeworks_py.Mesh, reference_particles: numpy.ndarray[numpy.float64[m, n]], landmarks: numpy.ndarray[numpy.float64[m, n]]) → bool

Invoked with: <shapeworks_py.MeshWarper object at 0x000002132F3BDE30>, PolyData (0x2132f485040)
N Cells: 17996
N Points: 9000
X Bounds: -5.189e+01, 3.424e+01
Y Bounds: -3.262e+01, 3.035e+01
Z Bounds: -1.396e+02, 9.510e+01
N Arrays: 0
, array([[ -2.94732, 5.13163, -75.2647 ],
[ -5.07009, -23.7317 , 26.3011 ],
[ 21.2718 , -5.59686, -30.3606 ],
…,
[ 9.99318, -25.7292 , 76.1548 ],
[ -29.259 , 13.1121 , 92.0959 ],
[ 21.045 , 15.0181 , -136.303 ]])’

Please let me know how to move forward with this. It difficult to debug it.
Thanks in advance
I

7

The generateWarp call takes a ShapeWorks Mesh class, instead of a vtkPolyData. No need to use sw.sw2vtkMesh

Try:

from shapeworks import *

shape_path = '/mean_geo/par_512.stl'
refrence_mesh = sw.Mesh(shape_path)
mean_particles = '/particles/mean_particles.particles'

particles = ParticleSystem([mean_particles]).ShapeAsPointSet(0)

warper = MeshWarper()
warper.generateWarp(reference_mesh, particles)
8

Thank you this was helpful.