| 1 |
""" |
| 2 |
.. _slice_example: |
| 3 |
|
| 4 |
Slicing |
| 5 |
~~~~~~~ |
| 6 |
|
| 7 |
Extract thin planar slices from a volume. |
| 8 |
""" |
| 9 |
|
| 10 |
import matplotlib.pyplot as plt |
| 11 |
import numpy as np |
| 12 |
|
| 13 |
# sphinx_gallery_thumbnail_number = 2 |
| 14 |
import pyvista as pv |
| 15 |
from pyvista import examples |
| 16 |
|
| 17 |
# %% |
| 18 |
# PyVista meshes have several slicing filters bound directly to all datasets. |
| 19 |
# These filters allow you to slice through a volumetric dataset to extract and |
| 20 |
# view sections through the volume of data. |
| 21 |
# |
| 22 |
# One of the most common slicing filters used in PyVista is the |
| 23 |
# :func:`pyvista.DataObjectFilters.slice_orthogonal` filter which creates three |
| 24 |
# orthogonal slices through the dataset parallel to the three Cartesian planes. |
| 25 |
# For example, let's slice through the sample geostatistical training image |
| 26 |
# volume. First, load up the volume and preview it: |
| 27 |
|
| 28 |
mesh = examples.load_channels() |
| 29 |
# define a categorical colormap |
| 30 |
cmap = plt.get_cmap('viridis', 4) |
| 31 |
|
| 32 |
mesh.plot(cmap=cmap) |
| 33 |
|
| 34 |
# %% |
| 35 |
# Note that this dataset is a 3D volume and there might be regions within this |
| 36 |
# volume that we would like to inspect. We can create slices through the mesh |
| 37 |
# to gain further insight about the internals of the volume. |
| 38 |
|
| 39 |
slices = mesh.slice_orthogonal() |
| 40 |
|
| 41 |
slices.plot(cmap=cmap) |
| 42 |
|
| 43 |
|
| 44 |
# %% |
| 45 |
# The orthogonal slices can be easily translated throughout the volume: |
| 46 |
|
| 47 |
slices = mesh.slice_orthogonal(x=20, y=20, z=30) |
| 48 |
slices.plot(cmap=cmap) |
| 49 |
# %% |
| 50 |
# We can also add just a single slice of the volume by specifying the origin |
| 51 |
# and normal of the slicing plane with the :func:`pyvista.DataObjectFilters.slice` |
| 52 |
# filter: |
| 53 |
|
| 54 |
# Single slice - origin defaults to the center of the mesh |
| 55 |
single_slice = mesh.slice(normal=[1, 1, 0]) |
| 56 |
|
| 57 |
pl = pv.Plotter() |
| 58 |
pl.add_mesh(mesh.outline(), color='k') |
| 59 |
pl.add_mesh(single_slice, cmap=cmap) |
| 60 |
pl.show() |
| 61 |
# %% |
| 62 |
# Adding slicing planes uniformly across an axial direction can also be |
| 63 |
# automated with the :func:`pyvista.DataObjectFilters.slice_along_axis` filter: |
| 64 |
|
| 65 |
slices = mesh.slice_along_axis(n=7, axis='y') |
| 66 |
|
| 67 |
slices.plot(cmap=cmap) |
| 68 |
|
| 69 |
|
| 70 |
# %% |
| 71 |
# Slice Along Line |
| 72 |
# ++++++++++++++++ |
| 73 |
# |
| 74 |
# We can also slice a dataset along a :func:`pyvista.Spline` or :func:`pyvista.Line` |
| 75 |
# using the :func:`pyvista.DataObjectFilters.slice_along_line` filter. |
| 76 |
# |
| 77 |
# First, define a line source through the dataset of interest. Please note |
| 78 |
# that this type of slicing is computationally expensive and might take a while |
| 79 |
# if there are a lot of points in the line - try to keep the resolution of |
| 80 |
# the line low. |
| 81 |
|
| 82 |
model = examples.load_channels() |
| 83 |
|
| 84 |
|
| 85 |
def path(y): |
| 86 |
"""Equation: x = a(y-h)^2 + k""" |
| 87 |
a = 110.0 / 160.0**2 |
| 88 |
x = a * y**2 + 0.0 |
| 89 |
return x, y |
| 90 |
|
| 91 |
|
| 92 |
x, y = path(np.arange(model.bounds.y_min, model.bounds.y_max, 15.0)) |
| 93 |
zo = np.linspace(9.0, 11.0, num=len(y)) |
| 94 |
points = np.c_[x, y, zo] |
| 95 |
spline = pv.Spline(points, n_points=15) |
| 96 |
spline |
| 97 |
|
| 98 |
|
| 99 |
# %% |
| 100 |
# Then run the filter |
| 101 |
slc = model.slice_along_line(spline) |
| 102 |
slc |
| 103 |
|
| 104 |
# %% |
| 105 |
|
| 106 |
pl = pv.Plotter() |
| 107 |
pl.add_mesh(slc, cmap=cmap) |
| 108 |
pl.add_mesh(model.outline()) |
| 109 |
pl.show(cpos=[1, -1, 1]) |
| 110 |
|
| 111 |
|
| 112 |
# %% |
| 113 |
# Multiple Slices in Vector Direction |
| 114 |
# +++++++++++++++++++++++++++++++++++ |
| 115 |
# |
| 116 |
# Slice a mesh along a vector direction perpendicularly. |
| 117 |
|
| 118 |
mesh = examples.download_brain() |
| 119 |
|
| 120 |
# Create vector |
| 121 |
vec = np.array([1.0, 2.0, 1.0]) |
| 122 |
# Normalize the vector |
| 123 |
normal = vec / np.linalg.norm(vec) |
| 124 |
|
| 125 |
# Make points along that vector for the extent of your slices |
| 126 |
a = mesh.center + normal * mesh.length / 3.0 |
| 127 |
b = mesh.center - normal * mesh.length / 3.0 |
| 128 |
|
| 129 |
# Define the line/points for the slices |
| 130 |
n_slices = 5 |
| 131 |
line = pv.Line(a, b, resolution=n_slices) |
| 132 |
|
| 133 |
# Generate all of the slices |
| 134 |
slices = pv.MultiBlock() |
| 135 |
for point in line.points: |
| 136 |
slices.append(mesh.slice(normal=normal, origin=point)) |
| 137 |
|
| 138 |
# %% |
| 139 |
|
| 140 |
pl = pv.Plotter() |
| 141 |
pl.add_mesh(mesh.outline(), color='k') |
| 142 |
pl.add_mesh(slices, opacity=0.75) |
| 143 |
pl.add_mesh(line, color='red', line_width=5) |
| 144 |
pl.show() |
| 145 |
|
| 146 |
|
| 147 |
# %% |
| 148 |
# Slice At Different Bearings |
| 149 |
# +++++++++++++++++++++++++++ |
| 150 |
# |
| 151 |
# From `pyvista-support#23 <https://github.com/pyvista/pyvista-support/issues/23>`_ |
| 152 |
# |
| 153 |
# An example of how to get many slices at different bearings all centered |
| 154 |
# around a user-chosen location. |
| 155 |
# |
| 156 |
# Create a point to orient slices around |
| 157 |
ranges = np.ptp(np.array(model.bounds).reshape(-1, 2), axis=1) |
| 158 |
point = np.array(model.center) - ranges * 0.25 |
| 159 |
|
| 160 |
# %% |
| 161 |
# Now generate a few normal vectors to rotate a slice around the z-axis. |
| 162 |
# Use equation for circle since its about the Z-axis. |
| 163 |
increment = np.pi / 6.0 |
| 164 |
# use a container to hold all the slices |
| 165 |
slices = pv.MultiBlock() # treat like a dictionary/list |
| 166 |
for theta in np.arange(0, np.pi, increment): |
| 167 |
normal = np.array([np.cos(theta), np.sin(theta), 0.0]).dot(np.pi / 2.0) |
| 168 |
name = f'Bearing: {np.rad2deg(theta):.2f}' |
| 169 |
slices[name] = model.slice(origin=point, normal=normal) |
| 170 |
slices |
| 171 |
|
| 172 |
# %% |
| 173 |
# And now display it. |
| 174 |
pl = pv.Plotter() |
| 175 |
pl.add_mesh(slices, cmap=cmap) |
| 176 |
pl.add_mesh(model.outline()) |
| 177 |
pl.show() |
| 178 |
|
| 179 |
# %% |
| 180 |
# Slice ImageData With Indexing |
| 181 |
# +++++++++++++++++++++++++++++ |
| 182 |
# Most slicing filters return :class:`~pyvista.PolyData` or |
| 183 |
# :class:`~pyvista.UnstructuredGrid`. For :class:`~pyvista.ImageData` inputs, however, |
| 184 |
# it's often desirable to return :class:`~pyvista.ImageData`. The |
| 185 |
# :meth:`~pyvista.ImageDataFilters.slice_index` filter supports this use case. |
| 186 |
# |
| 187 |
# Extract a single 2D slice from a 3D segmentation mask and plot it. Here we use |
| 188 |
# :func:`~pyvista.examples.examples.load_frog_tissues`. |
| 189 |
|
| 190 |
mask = examples.load_frog_tissues() |
| 191 |
sliced = mask.slice_index(k=50) |
| 192 |
colored = sliced.color_labels() |
| 193 |
colored.plot(cpos='xy', zoom='tight', lighting=False) |
| 194 |
|
| 195 |
# %% |
| 196 |
# Extract a 3D volume of interest instead and visualize it as a surface mesh. Here we |
| 197 |
# define indices to extract the frog's head. |
| 198 |
|
| 199 |
sliced = mask.slice_index(i=[300, 500], j=[110, 350], k=[0, 100]) |
| 200 |
surface = sliced.contour_labels() |
| 201 |
colored = surface.color_labels() |
| 202 |
|
| 203 |
cpos = pv.CameraPosition( |
| 204 |
position=(520.0, 461.0, -402.0), |
| 205 |
focal_point=(372.0, 243.0, 52.0), |
| 206 |
viewup=(-0.73, -0.50, -0.47), |
| 207 |
) |
| 208 |
colored.plot(cpos=cpos) |
| 209 |
|
| 210 |
# %% |
| 211 |
# .. tags:: filter |