Statistics using SimpleITK#

We can use SimpleITK for extracting features from 3D label images. For convenience reasons we use the napari-simpleitk-image-processing library.

import numpy as np
import pandas as pd
from skimage.data import cells3d
import stackview
from skimage import measure
from skimage import filters
import napari_simpleitk_image_processing as nsitk

# load image
image = cells3d()[:,1]
stackview.insight(image)
shape(60, 256, 256)
dtypeuint16
size7.5 MB
min0
max65535
 
# denoising
blurred_image = filters.gaussian(image, sigma=1)

# binarization
threshold = filters.threshold_otsu(blurred_image)
thresholded_image = blurred_image >= threshold

# labeling
label_image = measure.label(thresholded_image)

# visualization
stackview.insight(label_image)
shape(60, 256, 256)
dtypeint32
size15.0 MB
min0
max38

Measurements/ region properties#

We are now using the very handy function label_statistics which provides a table of features. Let us check first what we need to provide for this function:

nsitk.label_statistics?

Signature:
nsitk.label_statistics(
    intensity_image: 'napari.types.ImageData',
    label_image: 'napari.types.LabelsData',
    size: bool = True,
    intensity: bool = True,
    perimeter: bool = False,
    shape: bool = False,
    position: bool = False,
    moments: bool = False,
    napari_viewer: 'napari.Viewer' = None,
) -> 'pandas.DataFrame'
Docstring:
Measure intensity/shape/... statistics per label

Parameters
----------
intensity_image: ndarray, optional
    Can be None
label_image: ndarray
    Must be subsequently labeled
size: bool, optional
intensity: bool, optional
perimeter: bool, optional
shape: bool, optional
position: bool, optional
moments: bool, optional
napari_viewer: napari.Viewer, optional

Returns
-------
pandas DataFrame, in case napari_viewr is None, otherwise the DataFrame will be added to
the passed label_image's layer.features


See Also
--------
..[0] https://simpleitk.org/doxygen/latest/html/classitk_1_1simple_1_1LabelShapeStatisticsImageFilter
..[1] http://insightsoftwareconsortium.github.io/SimpleITK-Notebooks/Python_html/35_Segmentation_Shape_Analysis.html
File:      c:\users\haase\mambaforge\envs\tea2024\lib\site-packages\napari_simpleitk_image_processing\_simpleitk_image_processing.py
Type:      function

Feature categories which have are set to True are measured by default. In this case, the categories are size and intensity. But the rest might be also interesting to investigate. So we need them to True as well:

df = pd.DataFrame(nsitk.label_statistics(image, label_image,  
                              size=True,
                              shape=True, 
                              perimeter=True, 
                              position=True))

df
label maximum mean median minimum sigma sum variance bbox_0 bbox_1 ... equivalent_ellipsoid_diameter_0 equivalent_ellipsoid_diameter_1 equivalent_ellipsoid_diameter_2 equivalent_spherical_perimeter equivalent_spherical_radius number_of_pixels number_of_pixels_on_border perimeter perimeter_on_border perimeter_on_border_ratio
0 1 65535.0 22623.940316 19071.708984 4126.0 12083.017961 7.471330e+08 1.459993e+08 33 58 ... 24.610818 46.274120 55.381770 4977.797687 19.902778 33024 523 7186.004933 523.0 0.072780
1 2 47610.0 16500.850114 16255.751953 4837.0 4418.271669 6.720961e+08 1.952112e+07 87 133 ... 29.888411 49.597182 52.476713 5724.894744 21.344148 40731 0 6172.236982 0.0 0.000000
2 3 55766.0 15183.665955 14463.779297 4932.0 4399.126377 1.414799e+09 1.935231e+07 51 192 ... 28.534707 51.817058 120.357525 9939.460710 28.123960 93179 1049 15144.119180 1049.0 0.069268
3 4 65298.0 19238.187210 18559.716797 4458.0 6757.679557 4.229708e+08 4.566623e+07 228 195 ... 29.207059 30.203761 47.599101 3795.318904 17.378777 21986 1157 4286.533941 1157.0 0.269915
4 5 36134.0 13131.887959 12927.802734 5453.0 2797.962330 1.357010e+09 7.828593e+06 126 0 ... 27.940183 51.292200 137.713543 10649.307251 29.110908 103337 785 21514.657299 785.0 0.036487
5 6 34332.0 12687.766570 12415.810547 5833.0 2613.257575 4.404812e+08 6.829115e+06 86 48 ... 28.520444 40.164739 57.881841 5146.503243 20.237236 34717 0 8387.413862 0.0 0.000000
6 7 41113.0 15498.465402 15231.767578 5643.0 3879.807500 7.140453e+08 1.505291e+07 10 120 ... 30.029330 41.504116 70.599501 6215.014814 22.239046 46072 0 7250.531028 0.0 0.000000
7 8 42394.0 13376.216498 12927.802734 5880.0 3092.779705 4.901982e+08 9.565286e+06 27 158 ... 29.605400 44.627579 52.974339 5335.515795 20.605505 36647 0 7516.259868 0.0 0.000000
8 9 55624.0 15445.162122 14975.771484 3983.0 4369.205585 7.239456e+08 1.908996e+07 144 176 ... 27.890772 51.601386 62.200368 6286.753725 22.367029 46872 0 7703.261028 0.0 0.000000
9 10 45903.0 12296.049111 12159.814453 5169.0 2351.130239 3.923300e+08 5.527813e+06 0 0 ... 27.589543 42.667873 51.765670 4864.909579 19.675802 31907 910 8026.035532 926.0 0.115375
10 11 34617.0 12775.336218 12415.810547 6022.0 2542.662952 4.295962e+08 6.465135e+06 54 13 ... 29.709700 37.358972 57.862384 5038.209321 20.023186 33627 0 6971.419828 0.0 0.000000
11 12 39406.0 14132.811306 13695.791016 5596.0 3229.607077 5.340224e+08 1.043036e+07 137 73 ... 31.075148 46.908817 49.506772 5445.503874 20.816806 37786 0 6298.712627 0.0 0.000000
12 13 39738.0 14336.412193 13951.787109 4837.0 3421.796926 6.751303e+08 1.170869e+07 220 78 ... 26.939097 34.084241 97.951670 6306.410182 22.401968 47092 1782 9251.046757 1782.0 0.192627
13 14 32910.0 14267.065307 13951.787109 5027.0 3382.211567 6.359402e+08 1.143936e+07 160 115 ... 27.654190 49.888746 61.704873 6079.556006 21.995356 44574 0 7637.852741 0.0 0.000000
14 15 25560.0 12693.643645 12159.814453 5975.0 2712.361825 1.668833e+08 7.356907e+06 184 233 ... 22.385207 23.324345 48.090298 2693.819960 14.641290 13147 814 5034.421312 814.0 0.161687
15 16 23378.0 11729.898405 11647.822266 6971.0 1929.340114 2.794062e+07 3.722353e+06 0 165 ... 8.035226 15.888670 35.633420 862.538778 8.284845 2382 480 1533.882378 480.0 0.312931
16 17 21624.0 12051.440492 11903.818359 6829.0 1909.310342 4.506034e+07 3.645466e+06 61 0 ... 9.119146 21.321118 36.727611 1164.987059 9.628433 3739 646 1806.519762 646.0 0.357594
17 18 18067.0 12642.775056 12671.806641 7113.0 1887.721974 5.676606e+06 3.563494e+06 42 252 ... 3.421169 12.457051 20.121389 283.562210 4.750280 449 201 467.703703 201.0 0.429759
18 19 18494.0 13171.954545 13183.798828 7777.0 2080.440088 4.346745e+06 4.328231e+06 247 0 ... 4.063350 9.274316 16.724348 230.936946 4.286884 330 172 326.481063 189.0 0.578900
19 20 12993.0 11074.950000 10879.833984 9626.0 866.009327 2.214990e+05 7.499722e+05 0 41 ... 1.500034 3.803307 6.695281 35.631775 1.683890 20 8 41.625299 8.0 0.192191
20 21 15412.0 11762.400000 11647.822266 8536.0 1432.674746 5.293080e+05 2.052557e+06 0 54 ... 2.590166 4.244004 7.818268 61.182300 2.206521 45 26 68.982937 26.0 0.376905
21 22 25607.0 12046.942658 11903.818359 6212.0 1987.248259 5.882522e+07 3.949156e+06 191 0 ... 12.177307 22.230537 34.449813 1391.901835 10.524449 4883 531 2100.978140 531.0 0.252739
22 23 22667.0 14137.542265 13951.787109 8014.0 2564.794476 8.864239e+06 6.578171e+06 252 14 ... 3.931732 14.779420 20.607609 354.265737 5.309574 627 258 555.992667 258.0 0.464035
23 24 13278.0 12313.333333 12671.806641 11096.0 1112.726981 3.694000e+04 1.238161e+06 0 75 ... 0.000000 0.000000 0.000000 10.059235 0.894700 3 3 8.070601 3.0 0.371720
24 25 15933.0 11565.902778 11391.826172 8868.0 1472.300884 8.327450e+05 2.167670e+06 0 206 ... 3.295460 4.895938 8.522794 83.696209 2.580762 72 28 94.930218 28.0 0.294953
25 26 15933.0 11984.942029 11903.818359 8962.0 1716.507911 8.269610e+05 2.946399e+06 0 61 ... 2.810671 4.741784 9.887779 81.354863 2.544408 69 38 99.583883 38.0 0.381588
26 27 11333.0 11333.000000 11391.826172 11333.0 0.000000 1.133300e+04 0.000000e+00 6 51 ... 0.000000 0.000000 0.000000 4.835976 0.620350 1 0 3.004080 0.0 0.000000
27 28 22288.0 15056.166667 14975.771484 9247.0 3374.279528 1.806740e+05 1.138576e+07 163 72 ... 2.203916 2.521246 4.124510 25.347684 1.420248 12 0 24.542458 0.0 0.000000
28 29 16977.0 14581.750000 16255.751953 11144.0 2606.129234 5.832700e+04 6.791910e+06 164 179 ... 0.000000 0.000000 0.000000 12.185896 0.984745 4 0 10.499264 0.0 0.000000
29 30 37747.0 16443.567164 13951.787109 6639.0 7418.631290 1.101719e+06 5.503609e+07 75 163 ... 3.595463 4.844137 7.346914 79.775094 2.519583 67 0 82.973966 0.0 0.000000
30 31 16834.0 14415.500000 16767.744141 11997.0 3420.275501 2.883100e+04 1.169828e+07 184 186 ... 0.000000 0.000000 0.000000 7.676633 0.781593 2 0 5.641937 0.0 0.000000
31 32 22430.0 15296.428571 16511.748047 10764.0 3278.686644 2.141500e+05 1.074979e+07 69 159 ... 2.160799 3.186080 3.883813 28.091148 1.495133 14 0 26.197575 0.0 0.000000
32 33 26271.0 15447.663462 15487.763672 8346.0 3648.568557 1.606557e+06 1.331205e+07 199 90 ... 5.098551 5.807702 6.707854 106.948086 2.917303 104 0 105.057726 0.0 0.000000
33 34 18873.0 15003.800000 14719.775391 12661.0 2099.312522 1.500380e+05 4.407113e+06 128 72 ... 2.144512 2.397638 3.714405 22.446612 1.336505 10 0 20.224982 0.0 0.000000
34 35 22667.0 15211.444444 14719.775391 7540.0 4278.261961 4.107090e+05 1.830353e+07 42 38 ... 2.196427 3.738460 6.279941 43.523783 1.861051 27 0 48.509499 0.0 0.000000
35 36 16834.0 14494.666667 14975.771484 12282.0 1758.626130 8.696800e+04 3.092766e+06 0 132 ... 0.000000 0.000000 0.000000 15.968040 1.127252 6 6 14.624144 6.0 0.410280
36 37 19585.0 19300.500000 19583.701172 19016.0 402.343758 3.860100e+04 1.618805e+05 243 69 ... 0.000000 0.000000 0.000000 7.676633 0.781593 2 0 5.641937 0.0 0.000000
37 38 14321.0 12139.500000 14207.783203 9958.0 3085.106886 2.427900e+04 9.517884e+06 59 0 ... 0.000000 0.000000 0.000000 7.676633 0.781593 2 2 5.641937 2.0 0.354488

38 rows × 31 columns

These are all columns that are available:

print(df.keys())

Index(['label', 'maximum', 'mean', 'median', 'minimum', 'sigma', 'sum',
       'variance', 'bbox_0', 'bbox_1', 'bbox_2', 'bbox_3', 'bbox_4', 'bbox_5',
       'centroid_0', 'centroid_1', 'centroid_2', 'elongation',
       'feret_diameter', 'flatness', 'roundness',
       'equivalent_ellipsoid_diameter_0', 'equivalent_ellipsoid_diameter_1',
       'equivalent_ellipsoid_diameter_2', 'equivalent_spherical_perimeter',
       'equivalent_spherical_radius', 'number_of_pixels',
       'number_of_pixels_on_border', 'perimeter', 'perimeter_on_border',
       'perimeter_on_border_ratio'],
      dtype='object')

df.describe()
label maximum mean median minimum sigma sum variance bbox_0 bbox_1 ... equivalent_ellipsoid_diameter_0 equivalent_ellipsoid_diameter_1 equivalent_ellipsoid_diameter_2 equivalent_spherical_perimeter equivalent_spherical_radius number_of_pixels number_of_pixels_on_border perimeter perimeter_on_border perimeter_on_border_ratio
count 38.000000 38.000000 38.000000 38.000000 38.000000 38.000000 3.800000e+01 3.800000e+01 38.000000 38.000000 ... 38.000000 38.000000 38.000000 38.000000 38.000000 38.000000 38.000000 38.000000 38.000000 38.000000
mean 19.500000 29249.631579 14141.934960 14012.417763 7860.578947 3062.636210 2.624646e+08 1.371663e+07 96.631579 96.605263 ... 12.892735 20.459120 32.981898 2458.453815 10.268039 17830.736842 247.868421 3574.679397 248.736842 0.149901
std 11.113055 14853.985729 2419.956877 2179.065205 3112.768027 2110.473638 3.820512e+08 2.467465e+07 84.994569 74.168012 ... 12.762268 19.699178 34.788721 3095.863466 9.625117 26189.733760 426.475003 4893.955404 427.080579 0.178247
min 1.000000 11333.000000 11074.950000 10879.833984 3983.000000 0.000000 1.133300e+04 0.000000e+00 0.000000 0.000000 ... 0.000000 0.000000 0.000000 4.835976 0.620350 1.000000 0.000000 3.004080 0.000000 0.000000
25% 10.250000 17249.500000 12300.370167 12223.813477 5607.750000 1914.317785 2.159872e+05 3.664688e+06 14.250000 42.750000 ... 2.169706 3.324175 4.663368 29.976304 1.542322 15.500000 0.000000 30.054506 0.000000 0.000000
50% 19.500000 23022.500000 14135.176785 13951.787109 7042.000000 2662.809700 7.270422e+06 7.093011e+06 72.000000 74.000000 ... 4.580950 13.618236 20.364499 318.913973 5.029927 538.000000 2.500000 511.848185 2.500000 0.052877
75% 28.750000 38991.250000 15204.499822 14975.771484 9531.250000 3421.416570 4.777690e+08 1.170609e+07 163.750000 162.000000 ... 27.831627 41.169272 52.849932 5119.429763 20.183723 34444.500000 424.500000 7132.358657 424.500000 0.308437
max 38.000000 65535.000000 22623.940316 19583.701172 19016.000000 12083.017961 1.414799e+09 1.459993e+08 252.000000 252.000000 ... 31.075148 51.817058 137.713543 10649.307251 29.110908 103337.000000 1782.000000 21514.657299 1782.000000 0.578900

8 rows × 31 columns

Exercise#

Load data/blobs.tif, segment the objects and measure feret_diameter, flatness, roundness and elongation. Compare the measurements with scikit-image.