Datasets with 3 or more spatial dimensions

Most scikit-image functions are compatible with 3D datasets, i.e., images with 3 spatial dimensions (to be distinguished from 2D multichannel images, which are also arrays with three axes). skimage.data.cells3d() returns a 3D fluorescence microscopy image of cells. The returned dataset is a 3D multichannel image with dimensions provided in (z, c, y, x) order. Channel 0 contains cell membranes, while channel 1 contains nuclei.

The example below shows how to explore this dataset. This 3D image can be used to test the various functions of scikit-image.

from skimage import data
import plotly
import plotly.express as px
import numpy as np

img = data.cells3d()[20:]

# omit some slices that are partially empty
img = img[5:26]

upper_limit = 1.5 * np.percentile(img, q=99)
img = np.clip(img, 0, upper_limit)

fig = px.imshow(
    img,
    facet_col=1,
    animation_frame=0,
    binary_string=True,
    binary_format="jpg",
)
fig.layout.annotations[0]["text"] = "Cell membranes"
fig.layout.annotations[1]["text"] = "Nuclei"
plotly.io.show(fig)