Sholl analysis in Skan

Skan provides a function to perform Sholl analysis, which counts the number of processes crossing circular (2D) or spherical (3D) shells from a given center point. Commonly, the center point is the soma, or cell body, of a neuron, but the method can be used to compare general skeleton structures when a root or center point is defined.

%matplotlib inline
%config InlineBackend.figure_format='retina'

import matplotlib.pyplot as plt

import numpy as np
import zarr

neuron = np.asarray(zarr.open('../example-data/neuron.zarr.zip'))

fig, ax = plt.subplots()
ax.imshow(neuron, cmap='gray')
ax.scatter(57, 54)
ax.set_axis_off()
plt.show()

This is the skeletonized image of a neuron. The cell body, or soma, has been manually annotated by a researcher based on the source image. We can use the function skan.sholl_analysis to count the crossings of concentric circles, centered on the cell body, by the cell’s processes.

import pandas as pd
from skan import Skeleton, sholl_analysis

# make the skeleton object
skeleton = Skeleton(neuron)
# define the neuron center/soma
center = np.array([54, 57])
# define radii at which to measure crossings
radii = np.arange(4, 45, 4)
# perform sholl analysis
center, radii, counts = sholl_analysis(
        skeleton, center=center, shells=radii
        )
table = pd.DataFrame({'radius': radii, 'crossings': counts})
table

	radius	crossings
0	4	4
1	8	5
2	12	6
3	16	5
4	20	4
5	24	5
6	28	4
7	32	1
8	36	0
9	40	0
10	44	0

We can visualize this using functions from skan.draw and matplotlib.

from skan import draw

# make two subplots
fig, (ax0, ax1) = plt.subplots(nrows=1, ncols=2, figsize=(8, 4))

# draw the skeleton
draw.overlay_skeleton_2d_class(
        skeleton, skeleton_colormap='viridis_r', vmin=0, axes=ax0
        )
# draw the shells
draw.sholl_shells(center, radii, axes=ax0)
# fiddle with plot visual aspects
ax0.autoscale_view()
ax0.set_facecolor('black')
ax0.set_ylim(75, 20)
ax0.set_xlim(20, 80)
ax0.set_aspect('equal')

# in second subplot, plot the Sholl analysis
ax1.plot('radius', 'crossings', data=table)
ax1.set_xlabel('radius')
ax1.set_ylabel('crossings')

plt.show()