| 1 |
""" |
| 2 |
.. _volume_rendering_example: |
| 3 |
|
| 4 |
Volume Rendering |
| 5 |
~~~~~~~~~~~~~~~~ |
| 6 |
|
| 7 |
Volume render uniform mesh types like :class:`pyvista.ImageData` or 3D |
| 8 |
NumPy arrays. |
| 9 |
|
| 10 |
This also explores how to extract a volume of interest (VOI) from a |
| 11 |
:class:`pyvista.ImageData` using the |
| 12 |
:func:`pyvista.ImageDataFilters.extract_subset` filter. |
| 13 |
""" |
| 14 |
|
| 15 |
# sphinx_gallery_thumbnail_number = 3 |
| 16 |
import numpy as np |
| 17 |
import pyvista as pv |
| 18 |
from pyvista import examples |
| 19 |
|
| 20 |
# sphinx_gallery_start_ignore |
| 21 |
# volume rendering does not work in interactive plots currently |
| 22 |
PYVISTA_GALLERY_FORCE_STATIC_IN_DOCUMENT = True |
| 23 |
# sphinx_gallery_end_ignore |
| 24 |
|
| 25 |
# Download a volumetric dataset |
| 26 |
vol = examples.download_knee_full() |
| 27 |
vol |
| 28 |
|
| 29 |
# %% |
| 30 |
# Simple Volume Render |
| 31 |
# ++++++++++++++++++++ |
| 32 |
# |
| 33 |
|
| 34 |
# A nice camera position |
| 35 |
cpos = pv.CameraPosition( |
| 36 |
position=(-381.74, -46.02, 216.54), |
| 37 |
focal_point=(74.8305, 89.2905, 100.0), |
| 38 |
viewup=(0.23, 0.072, 0.97), |
| 39 |
) |
| 40 |
|
| 41 |
vol.plot(volume=True, cmap='bone', cpos=cpos) |
| 42 |
|
| 43 |
|
| 44 |
# %% |
| 45 |
# Opacity Mappings |
| 46 |
# ++++++++++++++++ |
| 47 |
# |
| 48 |
# Or use the :func:`pyvista.Plotter.add_volume` method like below. |
| 49 |
# Note that here we use a non-default opacity mapping to a sigmoid: |
| 50 |
|
| 51 |
pl = pv.Plotter() |
| 52 |
pl.add_volume(vol, cmap='bone', opacity='sigmoid') |
| 53 |
pl.camera_position = cpos |
| 54 |
pl.show() |
| 55 |
|
| 56 |
# %% |
| 57 |
# You can also use a custom opacity mapping |
| 58 |
opacity = [0, 0, 0, 0.1, 0.3, 0.6, 1] |
| 59 |
|
| 60 |
pl = pv.Plotter() |
| 61 |
pl.add_volume(vol, cmap='viridis', opacity=opacity) |
| 62 |
pl.camera_position = cpos |
| 63 |
pl.show() |
| 64 |
|
| 65 |
# %% |
| 66 |
# We can also use a shading technique when volume rendering with the ``shade`` |
| 67 |
# option |
| 68 |
pl = pv.Plotter(shape=(1, 2)) |
| 69 |
pl.add_volume(vol, cmap='viridis', opacity=opacity, shade=False) |
| 70 |
pl.add_text('No shading') |
| 71 |
pl.camera_position = cpos |
| 72 |
pl.subplot(0, 1) |
| 73 |
pl.add_volume(vol, cmap='viridis', opacity=opacity, shade=True) |
| 74 |
pl.add_text('Shading') |
| 75 |
pl.link_views() |
| 76 |
pl.show() |
| 77 |
|
| 78 |
# %% |
| 79 |
# Cool Volume Examples |
| 80 |
# ++++++++++++++++++++ |
| 81 |
# |
| 82 |
# Here are a few more cool volume rendering examples. |
| 83 |
|
| 84 |
|
| 85 |
# %% |
| 86 |
# Head Dataset |
| 87 |
# """""""""""" |
| 88 |
|
| 89 |
head = examples.download_head() |
| 90 |
|
| 91 |
pl = pv.Plotter() |
| 92 |
pl.add_volume(head, cmap='cool', opacity='sigmoid_6', show_scalar_bar=False) |
| 93 |
pl.camera_position = pv.CameraPosition( |
| 94 |
position=(-228.0, -418.0, -158.0), |
| 95 |
focal_point=(94.0, 122.0, 82.0), |
| 96 |
viewup=(-0.2, -0.3, 0.9), |
| 97 |
) |
| 98 |
pl.camera.zoom(1.5) |
| 99 |
pl.show() |
| 100 |
|
| 101 |
|
| 102 |
# %% |
| 103 |
# Bolt-Nut MultiBlock Dataset |
| 104 |
# """"""""""""""""""""""""""" |
| 105 |
# .. note:: |
| 106 |
# See how we set interpolation to ``'linear'`` here to smooth out scalars of |
| 107 |
# each individual cell to make a more appealing plot. Two actor are returned |
| 108 |
# by ``add_volume`` because ``bolt_nut`` is a :class:`pyvista.MultiBlock` |
| 109 |
# dataset. |
| 110 |
|
| 111 |
bolt_nut = examples.download_bolt_nut() |
| 112 |
|
| 113 |
pl = pv.Plotter() |
| 114 |
actors = pl.add_volume( |
| 115 |
bolt_nut, cmap='coolwarm', opacity='sigmoid_5', show_scalar_bar=False |
| 116 |
) |
| 117 |
actors[0].prop.interpolation_type = 'linear' |
| 118 |
actors[1].prop.interpolation_type = 'linear' |
| 119 |
pl.camera_position = pv.CameraPosition( |
| 120 |
position=(127.4, -68.3, 88.2), |
| 121 |
focal_point=(30.3, 54.3, 26.0), |
| 122 |
viewup=(-0.25, 0.28, 0.93), |
| 123 |
) |
| 124 |
cpos = pl.show(return_cpos=True) |
| 125 |
|
| 126 |
|
| 127 |
# %% |
| 128 |
# Frog Dataset |
| 129 |
# """""""""""" |
| 130 |
|
| 131 |
frog = examples.download_frog() |
| 132 |
|
| 133 |
pl = pv.Plotter() |
| 134 |
pl.add_volume(frog, cmap='viridis', opacity='sigmoid_6', show_scalar_bar=False) |
| 135 |
pl.camera_position = pv.CameraPosition( |
| 136 |
position=(929.0, 1067.0, -278.9), |
| 137 |
focal_point=(249.5, 234.5, 101.25), |
| 138 |
viewup=(-0.2048, -0.2632, -0.9427), |
| 139 |
) |
| 140 |
pl.camera.zoom(1.5) |
| 141 |
pl.show() |
| 142 |
|
| 143 |
|
| 144 |
# %% |
| 145 |
# Extracting a VOI |
| 146 |
# ++++++++++++++++ |
| 147 |
# |
| 148 |
# Use the :func:`pyvista.ImageDataFilters.extract_subset` filter to extract |
| 149 |
# a volume of interest/subset volume to volume render. This is ideal when |
| 150 |
# dealing with particularly large volumes and you want to volume render only |
| 151 |
# a specific region. |
| 152 |
|
| 153 |
# Load a particularly large volume |
| 154 |
large_vol = examples.download_damavand_volcano() |
| 155 |
large_vol |
| 156 |
|
| 157 |
# %% |
| 158 |
opacity = [0, 0.75, 0, 0.75, 1.0] |
| 159 |
clim = [0, 100] |
| 160 |
|
| 161 |
pl = pv.Plotter() |
| 162 |
pl.add_volume( |
| 163 |
large_vol, |
| 164 |
cmap='magma', |
| 165 |
clim=clim, |
| 166 |
opacity=opacity, |
| 167 |
opacity_unit_distance=6000, |
| 168 |
) |
| 169 |
pl.show() |
| 170 |
|
| 171 |
|
| 172 |
# %% |
| 173 |
# Woah, that's a big volume. We probably don't want to volume render the |
| 174 |
# whole thing. So let's extract a region of interest under the volcano. |
| 175 |
# |
| 176 |
# The region we will extract will be between nodes 175 and 200 on the x-axis, |
| 177 |
# between nodes 105 and 132 on the y-axis, and between nodes 98 and 170 on |
| 178 |
# the z-axis. |
| 179 |
|
| 180 |
voi = large_vol.extract_subset([175, 200, 105, 132, 98, 170]) |
| 181 |
|
| 182 |
pl = pv.Plotter() |
| 183 |
pl.add_mesh(large_vol.outline(), color='k') |
| 184 |
pl.add_mesh(voi, cmap='magma') |
| 185 |
pl.show() |
| 186 |
|
| 187 |
# %% |
| 188 |
# Ah, much better. Let's now volume render that region of interest. |
| 189 |
|
| 190 |
pl = pv.Plotter() |
| 191 |
pl.add_volume(voi, cmap='magma', clim=clim, opacity=opacity, opacity_unit_distance=2000) |
| 192 |
pl.camera_position = pv.CameraPosition( |
| 193 |
position=(531554.5542909054, 3944331.800171338, 26563.04809259223), |
| 194 |
focal_point=(599088.1433822059, 3982089.287834022, -11965.14728669936), |
| 195 |
viewup=(0.3738545892415734, 0.244312810377319, 0.8947312427698892), |
| 196 |
) |
| 197 |
pl.show() |
| 198 |
|
| 199 |
|
| 200 |
# %% |
| 201 |
# .. _volume_with_mask_example: |
| 202 |
# |
| 203 |
# Volume With Segmentation Mask |
| 204 |
# +++++++++++++++++++++++++++++ |
| 205 |
# Visualize a medical image with a corresponding binary segmentation mask. |
| 206 |
# |
| 207 |
# For this example, we use :func:`~pyvista.examples.downloads.download_whole_body_ct_male` |
| 208 |
# though :func:`~pyvista.examples.downloads.download_whole_body_ct_female`, or any |
| 209 |
# other dataset with a corresponding label or mask may be used. |
| 210 |
|
| 211 |
# %% |
| 212 |
# Load the dataset and get the ct image and a mask image. Here, a mask of the heart is |
| 213 |
# used. |
| 214 |
dataset = examples.download_whole_body_ct_male() |
| 215 |
ct_image = dataset['ct'] |
| 216 |
heart_mask = dataset['segmentations']['heart'] |
| 217 |
|
| 218 |
# %% |
| 219 |
# Use the segmentation mask to isolate the heart in the CT image. |
| 220 |
# |
| 221 |
# Initialize a new array and image with CT background values. Here, we set the scalar |
| 222 |
# values to ``-1000`` which typically corresponds to air (low density). |
| 223 |
|
| 224 |
heart_array = np.full_like(ct_image.active_scalars, -1000) |
| 225 |
|
| 226 |
# %% |
| 227 |
# Extract the intensities for the heart segment. We use heart mask's array |
| 228 |
# to mask the CT image to only extract the intensities of interest. |
| 229 |
ct_image_array = ct_image.active_scalars |
| 230 |
heart_mask_array = heart_mask.active_scalars |
| 231 |
heart_array[heart_mask_array == True] = ct_image_array[heart_mask_array == True] # noqa: E712 |
| 232 |
|
| 233 |
# %% |
| 234 |
# Add the masked array to the CT image as a new set of scalar values. |
| 235 |
ct_image['heart'] = heart_array |
| 236 |
|
| 237 |
# %% |
| 238 |
# Create the plot. |
| 239 |
# |
| 240 |
# For the CT image, the opacity is set to a sigmoid function to show the |
| 241 |
# subject's skeleton. Since different images have different intensity |
| 242 |
# distributions, you may need to experiment with different sigmoid functions. |
| 243 |
# See :func:`~pyvista.Plotter.add_volume` for details. |
| 244 |
|
| 245 |
pl = pv.Plotter() |
| 246 |
|
| 247 |
# Add the CT image. |
| 248 |
pl.add_volume( |
| 249 |
ct_image, |
| 250 |
scalars='NIFTI', |
| 251 |
cmap='bone', |
| 252 |
opacity='sigmoid_15', |
| 253 |
show_scalar_bar=False, |
| 254 |
) |
| 255 |
|
| 256 |
# Add masked CT image of the heart and use a contrasting color map. |
| 257 |
_ = pl.add_volume( |
| 258 |
ct_image, |
| 259 |
scalars='heart', |
| 260 |
cmap='gist_heat', |
| 261 |
opacity='linear', |
| 262 |
opacity_unit_distance=np.mean(ct_image.spacing), |
| 263 |
) |
| 264 |
|
| 265 |
# Orient the camera to provide a latero-anterior view. |
| 266 |
pl.view_yz() |
| 267 |
pl.camera.azimuth = 70 |
| 268 |
pl.camera.up = (0, 0, 1) |
| 269 |
pl.camera.zoom(1.5) |
| 270 |
pl.show() |
| 271 |
|
| 272 |
# %% |
| 273 |
# .. tags:: medical, plot |