{ "cells": [ { "cell_type": "markdown", "id": "3a0317a8-7eb0-49a1-9ea8-a7ae3724d210", "metadata": {}, "source": [ "# Clustering - a quick walkthrough\n", "\n", "The term _clustering_ stands for grouping objects according to their properties without providing any annotation. Algorithms in this field are also reffered to as unsupervised machine learning. The algorithms do receive input from the programmer though, e.g. by selecting specific measurements, which might render the term _unsupervised_ in this context a bit misleading.\n", "\n", "When clustering data retrieved from images, we use terms such as standard-scaling, dimensionality reduction and finally algorithms such as k-means clustering. This notebook is a quick walk through using these techniques before the methods are demonstrated in more detail in the following notebooks.\n", "\n", "See also\n", "* [Explorative image data science with napari (FocalPlane blog post)](https://focalplane.biologists.com/2022/05/23/explorative-image-data-science-with-napari/)" ] }, { "cell_type": "code", "execution_count": 1, "id": "ca31b334-2dfa-4dd8-8b0b-48b8e100986c", "metadata": {}, "outputs": [], "source": [ "import pyclesperanto_prototype as cle\n", "from napari_simpleitk_image_processing import label_statistics\n", "from sklearn.preprocessing import StandardScaler\n", "from sklearn.mixture import GaussianMixture\n", "from sklearn.cluster import KMeans\n", "import pandas as pd\n", "import seaborn as sns\n", "import seaborn\n", "import numpy as np\n", "import umap\n", "import matplotlib.pyplot as plt\n", "from skimage.data import human_mitosis" ] }, { "cell_type": "markdown", "id": "c2d27837-01b8-4202-89ef-7642a2851f14", "metadata": {}, "source": [ "First we start by loading a 3D dataset showing a Tribolium castaneum embryo undergoing gastrulation (curtesy: Daniela Vorkel, Myers lab, MPI-CBG / CSBD Dresden). The dataset shows dense nuclei marked with nuclei-GFP on the left forming the embryo and less dense nuclei, called serosa, on the right surrounding the embryo." ] }, { "cell_type": "code", "execution_count": 2, "id": "24e3f18b-6bcb-4ed4-aa9e-cc98090affcb", "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", "\n", "\n", "\n", "\n", "
\n", "\n", "\n", "cle._ image
\n", "\n", "\n", "\n", "\n", "\n", "
shape(116, 636, 354)
dtypefloat32
size99.6 MB
min0.0
max255.0
\n", "\n", "
" ], "text/plain": [ "cl.OCLArray([[[ 2., 1., 1., ..., 2., 2., 3.],\n", " [ 1., 2., 2., ..., 2., 2., 2.],\n", " [ 2., 2., 2., ..., 2., 2., 2.],\n", " ...,\n", " [ 8., 8., 7., ..., 10., 10., 10.],\n", " [ 6., 7., 7., ..., 9., 10., 10.],\n", " [ 7., 8., 8., ..., 9., 9., 9.]],\n", "\n", " [[ 2., 1., 1., ..., 1., 1., 2.],\n", " [ 1., 2., 2., ..., 1., 2., 2.],\n", " [ 1., 2., 2., ..., 1., 2., 2.],\n", " ...,\n", " [ 8., 7., 7., ..., 10., 10., 9.],\n", " [ 7., 8., 7., ..., 9., 10., 10.],\n", " [ 7., 8., 9., ..., 9., 9., 10.]],\n", "\n", " [[ 2., 1., 0., ..., 1., 1., 2.],\n", " [ 2., 2., 1., ..., 1., 2., 2.],\n", " [ 1., 2., 2., ..., 1., 2., 1.],\n", " ...,\n", " [ 8., 7., 7., ..., 10., 10., 9.],\n", " [ 8., 8., 8., ..., 8., 9., 10.],\n", " [ 7., 8., 9., ..., 9., 9., 10.]],\n", "\n", " ...,\n", "\n", " [[ 0., 0., 0., ..., 0., 0., 0.],\n", " [ 0., 0., 0., ..., 0., 0., 0.],\n", " [ 0., 0., 0., ..., 0., 0., 0.],\n", " ...,\n", " [ 3., 2., 3., ..., 3., 4., 3.],\n", " [ 2., 2., 3., ..., 3., 3., 3.],\n", " [ 2., 2., 3., ..., 3., 3., 3.]],\n", "\n", " [[ 0., 0., 0., ..., 0., 0., 0.],\n", " [ 0., 0., 0., ..., 0., 0., 0.],\n", " [ 0., 0., 0., ..., 0., 0., 0.],\n", " ...,\n", " [ 3., 2., 2., ..., 4., 3., 2.],\n", " [ 2., 3., 3., ..., 3., 3., 2.],\n", " [ 2., 3., 3., ..., 3., 3., 2.]],\n", "\n", " [[ 0., 0., 0., ..., 0., 0., 0.],\n", " [ 0., 0., 0., ..., 0., 0., 0.],\n", " [ 0., 0., 0., ..., 0., 0., 0.],\n", " ...,\n", " [ 3., 2., 2., ..., 3., 3., 2.],\n", " [ 2., 3., 3., ..., 2., 3., 2.],\n", " [ 2., 2., 3., ..., 3., 3., 2.]]], dtype=float32)" ] }, "execution_count": 2, "metadata": {}, "output_type": "execute_result" } ], "source": [ "image = cle.imread(\"../../data/Lund-25MB.tif\")\n", "image" ] }, { "cell_type": "markdown", "id": "c116668b-307b-4693-87fb-12c89122fb85", "metadata": {}, "source": [ "We segment the nuclei as shown in earlier sections using [top-hat-filtering for background removal](image-filtering:background_removal) and [Voronoi-Otsu-Labeling](image-segmentation:voronoi-otsu-labeling)." ] }, { "cell_type": "code", "execution_count": 3, "id": "b1a87314-cf1d-487c-99e3-0a90ffc7b2d5", "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", "\n", "\n", "\n", "\n", "
\n", "\n", "\n", "cle._ image
\n", "\n", "\n", "\n", "\n", "\n", "
shape(116, 636, 354)
dtypeuint32
size99.6 MB
min0.0
max1200.0
\n", "\n", "
" ], "text/plain": [ "cl.OCLArray([[[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]],\n", "\n", " [[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]],\n", "\n", " [[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]],\n", "\n", " ...,\n", "\n", " [[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]],\n", "\n", " [[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]],\n", "\n", " [[0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " ...,\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0],\n", " [0, 0, 0, ..., 0, 0, 0]]], dtype=uint32)" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "background_subtracted = cle.top_hat_box(image, radius_x=5, radius_y=5)\n", "nuclei_labels = cle.voronoi_otsu_labeling(background_subtracted, spot_sigma=1)\n", "nuclei_labels" ] }, { "cell_type": "markdown", "id": "fe181ab3-1813-4e42-b04d-af91c9cd217c", "metadata": {}, "source": [ "## Feature extraction\n", "We next measure properties such as intensity, size and shape from the labeled nuclei using [napari-SimpleITK-image-processing](https://www.napari-hub.org/plugins/napari-simpleitk-image-processing)." ] }, { "cell_type": "code", "execution_count": 4, "id": "59416a65-f55b-4c6c-9064-7f8635800dd6", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Index(['label', 'maximum', 'mean', 'median', 'minimum', 'sigma', 'sum',\n", " 'variance', 'elongation', 'feret_diameter', 'flatness', 'roundness',\n", " 'equivalent_ellipsoid_diameter_0', 'equivalent_ellipsoid_diameter_1',\n", " 'equivalent_ellipsoid_diameter_2', 'equivalent_spherical_perimeter',\n", " 'equivalent_spherical_radius', 'number_of_pixels',\n", " 'number_of_pixels_on_border', 'perimeter', 'perimeter_on_border',\n", " 'perimeter_on_border_ratio'],\n", " dtype='object')" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "statistics = label_statistics(image, nuclei_labels,\n", " intensity=True,\n", " perimeter=True,\n", " shape=True)\n", "statistics.keys()" ] }, { "cell_type": "code", "execution_count": 5, "id": "15492608-d9d1-41b7-960f-fc93fc55bcae", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
labelmaximummeanmedianminimumsigmasumvarianceelongationferet_diameter...equivalent_ellipsoid_diameter_0equivalent_ellipsoid_diameter_1equivalent_ellipsoid_diameter_2equivalent_spherical_perimeterequivalent_spherical_radiusnumber_of_pixelsnumber_of_pixels_on_borderperimeterperimeter_on_borderperimeter_on_border_ratio
01143.0117.489451117.04101693.09.48978627845.090.0560321.2286908.774964...6.5172007.5183609.237736185.2037133.83901623713191.79034913.00.067782
12113.083.05221982.17773465.09.69980831809.094.0862711.32509613.152946...7.2021788.75476411.600904255.0448984.50508938374311.44641474.00.237601
23130.0108.930403108.07617292.07.55705729738.057.1091091.56591112.884099...5.4492517.81681912.240444203.5131874.02430927374252.13096374.00.293498
34129.094.57699193.13476670.011.43311653436.0130.7161361.22702714.352700...7.66555710.71089913.142567330.5088475.12845656565396.76631065.00.163824
45149.0119.454545119.03320389.012.01795832850.0144.4313211.42982910.723805...6.1096277.75385511.086684204.5059374.0341132750234.6112780.00.000000
\n", "

5 rows × 22 columns

\n", "
" ], "text/plain": [ " label maximum mean median minimum sigma sum \\\n", "0 1 143.0 117.489451 117.041016 93.0 9.489786 27845.0 \n", "1 2 113.0 83.052219 82.177734 65.0 9.699808 31809.0 \n", "2 3 130.0 108.930403 108.076172 92.0 7.557057 29738.0 \n", "3 4 129.0 94.576991 93.134766 70.0 11.433116 53436.0 \n", "4 5 149.0 119.454545 119.033203 89.0 12.017958 32850.0 \n", "\n", " variance elongation feret_diameter ... \\\n", "0 90.056032 1.228690 8.774964 ... \n", "1 94.086271 1.325096 13.152946 ... \n", "2 57.109109 1.565911 12.884099 ... \n", "3 130.716136 1.227027 14.352700 ... \n", "4 144.431321 1.429829 10.723805 ... \n", "\n", " equivalent_ellipsoid_diameter_0 equivalent_ellipsoid_diameter_1 \\\n", "0 6.517200 7.518360 \n", "1 7.202178 8.754764 \n", "2 5.449251 7.816819 \n", "3 7.665557 10.710899 \n", "4 6.109627 7.753855 \n", "\n", " equivalent_ellipsoid_diameter_2 equivalent_spherical_perimeter \\\n", "0 9.237736 185.203713 \n", "1 11.600904 255.044898 \n", "2 12.240444 203.513187 \n", "3 13.142567 330.508847 \n", "4 11.086684 204.505937 \n", "\n", " equivalent_spherical_radius number_of_pixels number_of_pixels_on_border \\\n", "0 3.839016 237 13 \n", "1 4.505089 383 74 \n", "2 4.024309 273 74 \n", "3 5.128456 565 65 \n", "4 4.034113 275 0 \n", "\n", " perimeter perimeter_on_border perimeter_on_border_ratio \n", "0 191.790349 13.0 0.067782 \n", "1 311.446414 74.0 0.237601 \n", "2 252.130963 74.0 0.293498 \n", "3 396.766310 65.0 0.163824 \n", "4 234.611278 0.0 0.000000 \n", "\n", "[5 rows x 22 columns]" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "statistics.head()" ] }, { "cell_type": "markdown", "id": "13a391c4-0629-42b1-91c2-b93040279724", "metadata": {}, "source": [ "## Feature selection\n", "Selecting the right features for differentiating objects in an art. We will select some features here manually and will explain in the next sections how a good educated guess for selecting features can be made." ] }, { "cell_type": "code", "execution_count": 6, "id": "8ac53b54-2c43-41d4-b934-b6b6662afabc", "metadata": {}, "outputs": [], "source": [ "selected_statistics = statistics[\n", " [\n", " 'mean',\n", " 'variance',\n", " 'number_of_pixels',\n", " 'elongation',\n", " 'feret_diameter',\n", " ]\n", "].values" ] }, { "cell_type": "markdown", "id": "674971c6-f98f-4a76-a9e7-1532042bc43f", "metadata": {}, "source": [ "## Standard scaling\n", "The selected features need to be scaled so that all values range from -1 to 1. This is necessary for the following algorithms as they could misinterpret a perimeter of 5 microns less than a perimeter of 50 pixels, even if both might be exactly the same length physically ([Read more](machine_learning_basics.scaling))." ] }, { "cell_type": "code", "execution_count": 7, "id": "6217cd12-5d70-4b5a-8483-44cba39974bd", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(numpy.ndarray, (1200, 5))" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "scaled_statistics = StandardScaler().fit_transform(selected_statistics)\n", "\n", "type(scaled_statistics), scaled_statistics.shape" ] }, { "cell_type": "markdown", "id": "cb00d576-ed59-4c9a-a77f-bfc8f038902e", "metadata": {}, "source": [ "## Dimensionality reduction\n", "As the measured statistics are a large table with many columns, we cannot easily get a picture of the distribution of the data points. For a clustering algorithm it might also be challenging. Thus, we reduce the number of dimensions, e.g. using the [Uniform Manifold Approximation Projection (UMAP)](https://arxiv.org/abs/1802.03426)." ] }, { "cell_type": "code", "execution_count": 8, "id": "d8ce567c-3ba8-44b5-a237-4cfa729a1e40", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(numpy.ndarray, (1200, 2))" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "reducer = umap.UMAP(random_state=42)\n", "embedding = reducer.fit_transform(scaled_statistics)\n", "type(embedding), embedding.shape" ] }, { "cell_type": "code", "execution_count": 9, "id": "a5f36166-9a26-4cc2-960c-7084603fd614", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0, 0.5, 'UMAP2')" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "seaborn.scatterplot(x=embedding[:, 0], \n", " y=embedding[:, 1])\n", "\n", "plt.xlabel('UMAP1')\n", "plt.ylabel('UMAP2')" ] }, { "cell_type": "markdown", "id": "6feda872-e6fd-4e78-982c-c5a24137116d", "metadata": {}, "source": [ "## K-means clustering\n", "Eventually, we will automatically annotate all data points with a class. In the following we will separate data points into two classes. Looking at the UMAP visualized above, we need an algorithm that takes relationsships of datapoints locally into account. [K-means clustering](https://en.wikipedia.org/wiki/K-means_clustering) has the capability to group data points in a way so that the distance between the data points to the cluster centers is minimal. It basically splits the data points along a region in the plot where the density of the data points is lower than in cluster centers, e.g. in the plot above where UMAP1 is approximately 5. " ] }, { "cell_type": "code", "execution_count": 10, "id": "66cde5d9-a68a-48f8-8f10-2501ebafb720", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\rober\\miniconda3\\envs\\bio_39\\lib\\site-packages\\sklearn\\cluster\\_kmeans.py:1332: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=5.\n", " warnings.warn(\n" ] } ], "source": [ "num_classes = 2\n", "\n", "kmeans = KMeans(n_clusters=num_classes, random_state=42).fit(embedding)\n", "\n", "kmeans_prediction = kmeans.predict(embedding)" ] }, { "cell_type": "code", "execution_count": 11, "id": "856b239e-a5f2-4e0b-a9bf-099fd88413aa", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "seaborn.scatterplot(x=embedding[:, 0], \n", " y=embedding[:, 1],\n", " hue=kmeans_prediction)" ] }, { "cell_type": "markdown", "id": "ff24741c-cd64-40bb-b48d-7f54090c0a34", "metadata": {}, "source": [ "We can also project the cluster identifier (0 or 1) back into the 3D image space. We just need to make sure we cannot mix up these two classes with background." ] }, { "cell_type": "code", "execution_count": 12, "id": "018d3f87-8d69-499e-a222-c99fd09e937c", "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAKEAAAD8CAYAAAAfQcSfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAAAafElEQVR4nO1df+xe1Vn/PHYUprAM3JwUULqENSJRWBqcWWNQhutwkcakC0tMFiXhH3QlsZEyEqd/EJlpiP1nxmZDa2RjwLa6LNmQ4QhrwhjMsQmU7+goGU3rUBkZ+Acd9fGPe285PZxz73N+n/ve80m++b7vfc+957zv/bzPeX6/xMxoaCiJnym9gIaGRsKG4mgkbCiORsKG4mgkbCiORsKG4khGQiLaSkRrRHSIiHalmqdh/qAUfkIiWgfg+wCuAnAEwKMAPszMT0WfrGH2SCUJLwdwiJmfZebjAO4CcE2iuRpmjjcluu55AJ5Xnh8B8Bu2wUTUwjYLADOT6XgqEpomO4VoRHQ9gOsTzd8wI6Qi4REAFyjPzwdwVB3AzHsB7AWaJFw6UumEjwK4iIg2EtF6ANcC+FKiuRpmjiSSkJlfI6I/AXAfgHUA7mDmJ1PM1TB/JHHROC+ibceLgM0waRGThuJIZZg0OOLolrWTjzcc2DQ5bmzM3NAkYQVQCWh6bjpuGzNHNBLOGKtCxEbCjDi6Zc1IHH1rlW61q7IlN50wMnSSDUTRt1JXAkmknlSvrA1NEkaERJfzeR1wJ9WctuomCRNiIM6GA5uiSKk5STcXNBJGhI0kPtuv6RpTc+lkH7tGTYRuEZOEkBAn57VirscHuVO5VhoSC7cESpPMF42EjnBR+Ke2RxfEIFTM9cREI2Em1CKValmHiqYTOmJMksS6wSHx4alzS8aebTphI2EgXC3OqfGp9bqSFnJL5aoAMfSxIfTnc60YTvMUaDphJKSQKvo1Q0kyZpgMx0tY2I2EnvDZ1lQS2M6JedNNc9VomLTt2AMmaSLdIjcc2BQlbDd1jVi5hzm26EbCEdj0L50MOXWpEBJLr5tbcjbr2AIbsaYs2pq2O6klnmvNzUXjiNj+wNpqQ0oYIM1F4wjT1mR6LoFEP1O3/hpDaynRSDgCnYi1SDFAbgjZIDFscn0hGgmFqImAKmKQxMXSToFJPyER3QHggwBeYOZL+mPnAPgcgAsBPAfgQ8z84/61mwFcB+AEgI8y831JVl4ItgRV3QhIKaVioZYvlkQS/iOArdqxXQAeYOaLADzQPwcRXYyu+dGv9ud8su/aWj0k248tquCa3mU7Htv9Ems7TU3WSUnIzA8R0YXa4WsAXNE/3gfgQQA39cfvYuZXARwmokPourY+HGm9xeB6M1P7D12d1a5EyiklfcN272DmYwDAzMeI6Bf64+cB+KYy7kh/bDbw+fBt4Tj1cUiZZw5CSHydqdYTO3Y82aH15MAKO7WOVcVNffDS16VkjJmskCIdLOY1fa3jHxHRuQDQ/3+hPz7ZoXUAM+9l5s3MvNlzDdGR2gq06ZQmxErnz5HdEwpfEn4JwEf6xx8B8C/K8WuJ6HQi2gjgIgDfCltiepRwDqdOdhibdyoebhqfaj2AzEXzWXRGyNuI6AiAjwO4DcDdRHQdgB8C2A4AzPwkEd0N4CkArwG4gZlPRF91BqTWi2rQD/VttZTLRmIdf9jy0pWW8bcCuDVkUaWgGxK2YzGu3/A6WlIr4mcw1wxXgyWHpdxIGACTJZ3KIo2ZQlabRG6xYwNsW7CtwZFp7BiWmCkzhkWTUJoz6CNxJG3iGhE7tO3YgimCpHQGSzGsIWbHr7FrpcogX7QkHPsgpUq7ycUx5k9LQVjVCR661ZeQzouXhGPSzDeOHGOMK+a8tS9aEtaAKeklUQumIhlT0nEs6UKPmOhjYqAVOhXGmJ7laxCNOdhdCBRbB2yFThMwSQtfHcvnnFjSJaaUmpJ8sVxNjYRw/3kG6bjQJIUcDu8U57hi8YaJFCXdMKnPL1283yQhpmuMfUNitYXHYiNWalczTAQoLSlyrSN2xrSO1r0f0x+yNAISmwy1kNw1JBlrrW077iE1KHyswdodyS5egBTvZVGSMLeUCS27TLWW2nTVRZFwwNQN0ZMTXLOrU0k+mzPbRX1wga2M1ZbG5kvuxRgmthvhmywqjXRMzeUL10hIaDpai5gEYkrHCwltma5f23YHpGkzEguLIKENtmC9urVIM6h1Mqe+4TnJlHquRWzHqm6j6zlSfcmlGMpXd5sjXN7bYtsF+2QN63A9L0SPjJUlncOZ7aozLlYnnLJs9+w/jD37D3vftCnDZuyY6bVYlnVsC13N3h4Qi+iLctHcs3M9jmLt5GMdU8XvNrhu7a6o2cfXYscTsJFCJ+D23ccBhNf2ThF37HXJazHWEIJQh7t37JiILgDwTwB+EcD/AdjLzHtWpWXwQECg25rV475ZyLabFVpYJYHNmazDR7qmksISnfA1AH/GzL8C4D0AbujbAs+2ZfD23cexffdx7Ni28eQxXTres3N9lO1VJbYvXFLKTLpb7ZA0RDoG4Fj/+GUiOoiu+2r1LYN1qaATbc/Ow7hHO0eVjOo1fKRADAKq63BFynh1zG3fyTrue1dfBuARaC2DAagtg59XTjO2DCai64noMSJ6zGPdYrhEClQCque4lnHWZDzYMmRCKudiS1mxdUxEZwL4PIAbmfknREYdExC2DGbmvQD29tdObh1JpNIgKdVteoBEh1KPq/PlzKBxSbYYM6BySl6RJCSi09AR8E5m/kJ/OLhlcA4MZNixbeNJcu3YtvGkXjh2zoAa9Ss9zGhCiJSTOO+lPtIpTJKQOpH3aQAHmfl25aVZtgzesW2jSCqG+uZUaar6EXMTOnTbta05pmSf9BMS0RYA3wDwH+hcNADwMXR64d0Afgl9y2BmfrE/5xYAf4zOsr6Rmb8yMUd2Z+WYwaJiIFMoKWNn2qRyYKdMQ1ts7HgKR7esiUiojgfGb8qe/Yet502d6zpXSsT+4iw2dhyKqa17iD0P4/T/Y1uwRK8bxpWA1C8Zur7Fk9AUwtMNFpVkwzl79h82fviqBNyz/zDu2bn+5J8v5lCBF4JFJTCYMJBmINkUWTYc2IQ9O98oHfXrjJ2vPh7bcn0zvGMTpyW1ZoI0uqFb13ryg+oSMl3X5IOMAan+ljIjZ+pL0HTCCUjJoRsdGw5sOrk96wR0ue6S0SShgpix3hLkc60SNI0L2dKbJKwMJos5NVwr6nzLHWI73RdvmKiQGhcxYNq6bYhpcIQS1LcWZwxNEhrgupWqcWnbc5tknCJ8ivqTMZSoTW46oYIQCahLURORbdcfI71Pp4XUJPKVzE0nHIHujI51TR26hByOjcElTzFnVnXWBIYcKCkJc+h/OSzllK1IJJJPMqZJQgPGCBiTOFM6YAxrM3Uav/7YNsYHi5aEUzra8OGGxH1tMKWIAelKNU3XdpFwA0JKUpskFEI1MELJN+iAtnKBHLpbqCST6KShX5xF+gnHtuGYOqIa4pNkdJfKlpmaVyL9Qta+OEmYK3phwpSemap/jA028sRYh8s1FkVCH0MkdEtW8w7V+V1CbJLEV9M5KlzmMp3vAtdzF7Edl5R+QJ8Eq+Qg6oRPsQ1L24HEmCcUi5KEY8hF1O27j2czSnwq7aTnjElWV2IuwkVTWhIOMLUYkSJnAX0qLNpF4+p4HupMbMXxOdagwzfVf+x1afcu0zjpMQlWXhK6SkFpcbwJajs59RoqAU2OXdeUe2kfxNAxtrl838Mif9vOh4BHt6wBu48HWcU2AgJxi95jbtE2ErVqu5nhnp3rgQOvP5duwSmq43ybIbnOEwOSNiBnAHgIwOnoSHsvM388ZqfW2NtxiCGi6oExwnaxkbr9R0rJF7Idvwrgd5j5lb471wEi+gqAP0DXqfU2ItqFrlPrTVqn1g0AvkZE72LmE1HeSQbESFhIlb41dwvZhEnrmDu80j89rf9jdB1Z9/XH9wHY1j8+2amVmQ8DGDq1Vo3BGo6VMVNjOzkJSqxb2p9wHRE9jq4H4f3MXG2nVlMzIhfEdMvMEVOWspSk0WPHzHyCmS9F1/DyciK6ZGS4uFMrM29m5s2ilTrAVycM7RkzYCCyxFfnilD/ng1TseWxlDDbc+n8Ts5qZn4JXYP0rai0U6stf28K9+xcH0UKmlrCjTl2fZMSpF285qAWSH7H5O0AfsrMLxHRmwG8D8An8Hqn1tvwxk6tnyGi29EZJtk6tcYIz4XohQMBcyUP+MDXCh4bHxpDlkjCcwF8nYi+B+BRdDrhl9GR7yoiegbAVf1zMPOT6Dq4PgXgqwBuWJplrGOs45avtZuyiH143bemxDXJdSXCdrVIwDnANVynjguth1l0AoMEvhJpTgRU4fN+JXXPPmlqLWzXo1S6V45qu5Dr6+MHfTfmOmdPwlLk8f3BnRphWrePISJ9XcesSRiDgD6pW9JOWqFSxzfFSzpeP8fn2ur7bF25PDCQydU/GNP9Yit4yiVVY1w79POYtSR0hSr1TilyD7yuKVQYcnN1AqYmpE8NSkzMloS+W/HYVirZmsd+D2+Dh6WsO48l/jtX53AMQ8K1HsblnNn6CUP1Qd+urK6/CD8Gk2UcI/3fNkcOY2nM2m9+Qg2+PQmlPQJ9YVLyQxV/H+QqSwUWSkJ1S46VuqXeNOnNs9VyjIX5fOGznbpgeP8+9Smz1AlDt+LBkIjReSsUPuRy0fFytpobXnOdc5GScM6I0StGvZbrdabqjX1Iv0gShsR7bVJY9fW5OIpDdK9UTYtsNcaS8T6Y5XYcipShvpzGQ4nQoC12HLKWRZLQFy4SNJVrRGqdx4rvpjCSdMzST1hT0oIJpi3MJ5Yb2n8mt6Sc+gIssg2IBIPzeYrY+o9pm46HotaSAAlC1r44w2THto2ndNxyddFIpLCvvhTDF2hLiHCFNLU/xhdncSQ0QUIsVxXAlwyxoyMhbhiX0s5mmDhgSDQ4umXNqa5k++7j2IE86fwuoUE9vmxzr5iIErv/zGLyCWMZJTG2LB05dDqpdALCW7yZ1IoU1vLsrONYJHTVCV1+/iF1pkvsTJtYmFqDzTqeHQkHuJDR9DOw6gcmIaJ6zlgW9AApOULcOWPXKkFEXxLObjsOgdoiJMQqlhR/m8ZJCBdCHtdwWy1YBAnVX1wf/kxk9DU6fOLG+vm+54bGn2PMpadxJcui6dvDfYeIvtw/P4eI7ieiZ/r/ZytjbyaiQ0S0RkTvd1pRIpi2b/3DmtriB8JK0+tdpFysgiOTa8jXTeND7NRZNDsAHFSe70LXqfUiAA/0z6F1at0K4JNEtM55ZYGQGBI+Ro7rNh7LeSydS8dUhy6TdMvd1UvaJPN8AL8H4FPK4aKdWkMLllLMGxOu2dkS6Vxr6ajUWf23AP4cwFnKsVM6tRKR2qn1m8o4a6dWANe7LnhAjJ+LjdkeWEdOSzVGxoxOVpMnwLXKT4pJSUhEHwTwAjN/W3jN4p1aU8FXCubSrXww5oBOXdQ1QCIJ3wvg94noagBnAHgLEf0z+k6tvRSsplOrFCc/4ANuPkdJqCuFHlXaB6gj5nuUdO+/mZnPZ+YL0Rkc/8bMf4jXO7UCb+zUei0RnU5EG5GxU6uKKZ1xwJ79h8UVd1KFPTVJJAXyS/ETFuvUGivrxfVGDY3VfSzkWGtwud5ciOiURcPMD6JrnA5m/h8AV1rG3Qrg1sC1JcPgrE5Z7ukSA5akQsXyI5qu57PVx5T2i0vlAvy7cekYI0/sNCkfxCRuyvexiLCdjiF0N3ywIT8fIdHPphC7ECqWczzXdr6ILJoBvk2QTFBJOxWpyFHkFAqbVB+OT2UQSda7Mlk0vuRTLWJff5/PeTHSsnJgzBEtMXZC1jsrSRirPTBw6odWy09H+ErC1BI0lkqxEkmtsePBprCdz+8dm4g9ZbBMjXOB6zVDM7l9E3Fb3bEBsX5McQfm45PTIdHpUqegzU4nDIG+ferP1bqTKYKqtctAHaE0IK8OGus9L0oSmrZzNe0rlvN6SnLUUmpZy3yz0Qlj6YN63Ng39zBHXmEsg2OM9DELraYwe8PEp7pOAl9y2xqoS7NsfEs7XSExWnIRcWX8hBKk7tpl+2V3aSq9+jhGxkvoNaRfiFTZOStJQl9IJGgqY8T35rpkzUit3Km66thkXFkSqtIwhWT0aYfhStoYcWCXa6jjcuYkrrR1PEU+/XWftm+63uZKxBqybXTY9MhUpFxZSWiDWggfihg3JfQG+1bTSeczfclil7DOxjoG6msTHMOqTBHGC50vVSx69mG7o1vWgn+NM3R+FT46YYy5SxXRp8TitmNX6B28VKjbku+Ny1WcXjNmIwkBvwyX0Pl2CMa5FDLllKDq/OqXReIsV89LrTKspCSM9aOJwHTvwRyujClpa1vDmATXz62+F02NUImmGw4hNSNTKLVlSvIT51juCcxsO9YxxG+PblnDBi0RIUa29PaJbahmHc6WfKqilvYksyGh7UMdjsVM/Y+JOZZ8Nus4AqTJq2NjADfLNZd+qCPEuk7lfHaFSBIS0XMAXgZwAsBrzLyZiM4B8DkAFwJ4DsCHmPnH/fibAVzXj/8oM98XfeUaQqTY9t3HT9l6Q29IzlSt0HNrgIsk/G1mvlRp5Za9U6vLh62S0iSdhvR8k1RU063GrhEC1+vNrcmRC0J0wmsAXNE/3oeuR81NUDq1AjhMREOn1ocD5pqE8QaNWMi6m0I6R8yUfJ/4bQqUTqKQkpAB/Gsf4/17Zt6LAp1ac0mCsUJw13Om5olJbB+n8pS0z0FMKQnfy8xHe6LdT0RPj4wVd2oFsBfwawMCuEkTWzLq2I2TXj9Et3MZ75McYSqsKi35dIhIyMxH+/8vENEX0W2vVXdqlYSbYkjWHNI5xNUyJulq0TElPat/jojOGh4D+F0AT6BAp1bX7OAp0knjp7bnNUMaoza5aUKTMlwhkYTvAPBFIhrGf4aZv0pEjwK4m4iuA/BDANuBrlMrEQ2dWl9Dgk6tIRjTwUL0upD1hK5h7Nyx9Y0RNedWPaukVsB800Lram3n51DUQ40Jn/XENIZcsNIln8NWYopa2IgpqQ2JGSuOtZXHLClIOYcLZi8JpbBtS/pxX8liO9clnd5VCkrPcUFKy3llJKHvh2Mj75jx4eqecR0/oHTstjRmk0UjwZSR4OP3M0mGUjqVvg5f6P5D9bol3tdKkRCwO6LH0tv1cRKpKVULTDe7BIlrdjfNTiccIPkQYxoSYzrlMD5X5oz0GmN1IjpyfClm35XLhJQfrM0ACDUMXFxM0i+CDinpUhs5OmZfd+yDsQ95SpqMObRTJR1MXTdkC53yp5bE7KzjEJjyBH0QGtUIsYb1BAvJ2lR9uDYCAiu8HUv9gupxX4ToeS5JFlNSMmTukjrhSm/HNoTGe8fgSoapsdJr+ca9a8Cst2PXGxgzccEVq5yeH4rZS0KbVJMaJS7nmcb5pIPVIoFqwawl4QCJwi0xSnxqPnRS1UgwU2JHTVgJEroglCRjuXkurh7fuaeIVDPZbFgcCXW4SjFfAyDUNSMJGZrGzIGQiyGhLeJhG6c/dhnjsv2pY1MQpnZVAVgxEtq2Q+mHPxBBco5tTGiiQCwiStZWC2ZvHeswhad80rdqS9WSqgqS0F9tEnGlJKEJvommvhLNtSQgxnapbuexnN85sXKSEHA3MFxzCH1vtM0HmYsYNRIQWFESqpASzHReCYRmupikYa3kG7DI7XjspugGh6tRE6L8S90w+lxTeYi1Y+VJmMNKjJE6n3pNNWPlt+MY8NnOXLdFSWaPj7un9q0YEEpCInorEd1LRE8T0UEi+k0iOoeI7ieiZ/r/ZyvjbyaiQ0S0RkTvT7d8GaRujzGDIkc4zDVhdQ6OaAlESa1EtA/AN5j5U0S0HsDPAvgYgBeZ+TYi2gXgbGa+qe/U+ll0nbs2APgagHeN9aPxTWqNAYnOKMmsyZ0gakLsZN3Y8C50IqK3APgugHeyMpiI1gBcobSGe5CZN/X9qsHMf92Puw/AXzLzwyNzFE3vnrp5td9cFTV8GWwIyax+J4D/AvAPRPTrAL4NYAcCO7XWjjF9KnYKvzomlDi1EU8CiU74JgDvBvB3zHwZgP9F3yTdAlGnViK6nogeI6LHRCstiFj1J5IxJku79nzAUEhIeATAEWZ+pH9+LzpS/qjfhuHTqZWZ9zLzZuXXAIphygiQuGBsZEkhmVaNkJMkZOb/BPA8EQ2f5pXoGmBm79SaElPuEx0uJFAd4DGTaleFiFI/4Z8CuLO3jJ8F8EfoCDy7Tq1jkMZ8Q6+v638uX4BVIZ6KWdcd54TUQnapI7ZdQ7qOuRkhre44AVKG2sYs87lEQqRY+dhxLKS+6S7x51UiINAkoRNcDZRVI0sqNBIGwlRCEMPAWRKBm2FSALGiI3PDSjbJbJgXVqZ7f8PqoRad8BUAtXhh3wbgv0svAqu3jl+2vVALCddqiCEDABE9VsNalrSOth03FEcjYUNx1ELCvaUXoKCWtSxmHVW4aBqWjVokYcOCUZyERLS1Lw091FftpZzrDiJ6gYieUI5lL10loguI6Ot9+eyTRLSjxFqI6Awi+hYRfbdfx1+VWAeYudgfgHUAfoCumGo9uqq+ixPO91voShOeUI79DYBd/eNdAD7RP764X8/pADb261wXaR3nAnh3//gsAN/v58u6FnT1QGf2j08D8AiA9+ReR2lJeDmAQ8z8LDMfB3AXgGtSTcbMDwF4UTt8DYB9/eN9ALYpx+9i5leZ+TCAQ/16Y6zjGDP/e//4ZQAH0VUkZl0Ld3ilf3pa/8e511GahOcBeF55XqI89JTSVQBq6WrytRHRhQAuQyeFsq+FiNYR0ePoCtXu566gLes6SpNQVB5aCMnXRkRnAvg8gBuZ+Scl1sLMJ5j5UnRVkZcT0SW511GahKLy0MQIKl31BRGdho6AdzLzF0quBQCY+SUADwLYmnsdpUn4KICLiGhjX8l3LbqS0ZzIXrpKRATg0wAOMvPtpdZCRG8norf2j98M4H0Ans69jqLWcW9xXY3OOvwBgFsSz/VZAMcA/BTdt/o6AD8P4AEAz/T/z1HG39Kvaw3AByKuYwu6bex7AB7v/67OvRYAvwbgO/06ngDwF/3xrOtoEZOG4ii9HTc0NBI2lEcjYUNxNBI2FEcjYUNxNBI2FEcjYUNxNBI2FMf/A45gLHR+si5CAAAAAElFTkSuQmCC\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "# suffix [0] represents background\n", "# we add 1 to the measurement so that there is no \n", "# cluster with ID 0, which corresponds to background.\n", "cluster_id = [0] + (kmeans_prediction + 1).tolist()\n", "\n", "kmeans_prediction_map = cle.replace_intensities(nuclei_labels, cluster_id)\n", "\n", "cle.imshow(kmeans_prediction_map, labels=True)" ] }, { "cell_type": "markdown", "id": "72684bec-3e3f-4c6f-926f-04373f00b4bd", "metadata": {}, "source": [ "## Interpreting clustering results\n", "For technical reasons it is illegal to claim that k-means clustering can differentiate serosa and embryo. The algorithm was not informed about what to differentiate and thus, just differentiates objects according to their properties. It is now the task of the scientists to figure out which parameters influenced the decision making. For example, if shape-based parameters dominated the decision making, a follow-up questions would be why so. It might be possible that nuclei within the embryo are differently shaped than nuclei within the serosa. But it is also possible that the nuclei segmentation algorithm caused objects to be segmented wrongly. Thus, clustering is just a tool for generating new hypotheses which need to be investigated further, before an algorithm can be designed that differentiates nuclei in serosa and embryo. Such an algorithm then also needs to be validated, e.g. by processing new, unseen datasets, and measuring the quality of the clustering quantitatively, e.g. by providing ground-truth annotations and comparing with them." ] }, { "cell_type": "code", "execution_count": null, "id": "ff504759-7157-4694-ada8-4f2ff038ae8a", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.9.13" } }, "nbformat": 4, "nbformat_minor": 5 }