A plugin to use with napari for clustering objects according to their properties
A plugin to use with napari for clustering objects according to their properties.
- Troubleshooting installation
For clustering objects according to their properties, the starting point is a grey-value image and a label image representing a segmentation of objects. For segmenting objects, you can for example use the Voronoi-Otsu-labelling approach in the napari plugin napari-segment-blobs-and-things-with-membranes.
The first step is deriving measurements from the labelled image and the corresponding pixels in the grey-value image.
You can use the menu
Tools > Measurement > Measure intensity, shape and neighbor counts (ncp) for that.
Just select the image, the corresponding label image and the measurements to analyse and click on
A table with the measurements will open:
Afterwards, you can save and/or close the measurement table. Also, close the Measure widget. If you are uploading your own measurements make sure that there is a column that specifies the which measurement belongs to which label by adding a column with the name "label". If you don't specify this column it will be assumed that measurements start at 1 and each column describes the next label.
If you have 3D time-lapse data this will automatically be detected. In case you have 2D time-lapse data you need to
convert it into a suitable shape using the function:
Tools > Utilities > Convert 3D stack to 2D time-lapse (time-slicer),
which can be found in the napari time slicer.
Note that tables for time-lapse data will include an additional column named "frame", which indicates which slice in
time the given row refers to. If you want to import your own csv files for time-lapse data make sure to include this column!
If you have tracking data where each column specifies measurements for a track instead of a label at a specific time point,
this column must not be added.
Once measurements were made, these measurements were saved in the
properties of the labels layer which was analysed.
You can then plot these measurements using the menu
Tools > Measurement > Plot measurement (ncp).
In this widget, you can select the labels layer which was analysed and the measurements which should be plotted
on the X- and Y-axis. If you cannot see any options in axes selection boxes, but you have performed measurements, click
Update Axes/Clustering Selection Boxes to refresh them. Click on
Run to draw the data points in the plot area.
You can also manually select a region in the plot. To use lasso (freehand) tool use left mouse click, and to use a rectangle - right click. The resulting manual clustering will also be visualized in the original image. To optimize visualization in the image, turn off the visibility of the analysed labels layer.
Hold down the SHIFT key while annotating regions in the plot to manually select multiple clusters.
When you plot your time-lapse datasets you will notice that the plots look slightly different. Datapoints of the current time frame are highlighted in white and you can see the datapoints move through the plot if you press play:
You can also manually select groups using the lasso tool and plot a measurement per frame and see how the group behaves in time. Furthermore, you could also select a group in time and see where the datapoints lie in a different feature space:
Dimensionality reduction: UMAP, t-SNE or PCA
For getting more insights into your data, you can reduce the dimensionality of the measurements, e.g.
using the UMAP algorithm, t-SNE
To apply it to your data use the menu
Tools > Measurement > Dimensionality reduction (ncp).
Select the label image that was analysed and in the list below, select all measurements that should be
dimensionality reduced. By default, all measurements are selected in the box. If you cannot see any measurements, but
you have performed them, click on
Update Measurements to refresh the box. You can read more about parameters of both
algorithms by hovering over question marks or by clicking on them. When you are done with the selection, click on
and after a moment, the table of measurements will re-appear with two additional columns representing the reduced
dimensions of the dataset. These columns are automatically saved in the
properties of the labels layer so there is no
need to save them for usage in other widgets unless you wish to do so.
Afterwards, you can again save and/or close the table. Also, close the Dimensionality Reduction widget.
If manual clustering, as shown above, is not an option, you can automatically cluster your data, using these implemented algorithms:
- k-means clustering (KMEANS)
- Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN)
- Gaussian Mixture Model (GMM)
- Mean Shift (MS)
- Agglomerative clustering (AC)
Therefore, click the menu
Tools > Measurement > Clustering (ncp),
again, select the analysed labels layer.
This time select the measurements for clustering, e.g. select only the
Select the clustering method
KMeans and click on
The table of measurements will reappear with an additional column
ALGORITHM_NAME_CLUSTERING_ID containing the cluster
ID of each datapoint.
Afterwards, you can again save and/or close the table. Also, close the clustering widget.
Plotting clustering results
Return to the Plotter widget using the menu
Tools > Measurement > Plot measurement (ncp).
UMAP_1 as X- and Y-axis and the
Clustering, and click on
Example of k-means clustering results:
- Get a python environment, e.g. via mini-conda. If you never used python/conda environments before, please follow the instructions here first. It is recommended to install python 3.9 to your new conda environment from the start. The plugin is not yet supported with Python 3.10. Create a new environment, for example, like this:
conda create --name ncp-env python=3.9
- Activate the new environment via conda:
conda activate ncp-env
- Install pyopencl, e.g. via conda:
conda install -c conda-forge pyopencl
python -m pip install "napari[all]"
Afterwards, you can install
napari-clusters-plotter via pip:
pip install napari-clusters-plotter
- If the plugin does not appear in napari 'Plugins' menu, and in 'Plugin errors...' you can see such an error:
ImportError: DLL load failed while importing _cl
Try downloading and installing a pyopencl with a lower cl version, e.g. cl12 : pyopencl=2020.1. However, in this case, you will need an environment with a lower python version (python=3.8).
Error: Could not build wheels for hdbscan which use PEP 517 and cannot be installed directly
Install hdbscan via conda before installing the plugin:
conda install -c conda-forge hdbscan
ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject
Similar to the above-described error, this error can occur when importing hdbscan through pip or in the wrong order. This can be fixed by installing packages separately through conda and in the following order:
conda install -c conda-forge napari pyopencl hdbscan pip install napari-clusters-plotter
WARNING: No ICDs were foundor
LogicError: clGetPlatformIDs failed: PLATFORM_NOT_FOUND_KHR
Make your system-wide implementation visible by installing either of the following conda packages:
conda install -c conda-forge ocl-icd-system conda install -c conda-forge ocl_icd_wrapper_apple
Contributions are very welcome. Tests can be run with pytest, please ensure the coverage at least stays the same before you submit a pull request.
Distributed under the terms of the BSD-3 license, "napari-clusters-plotter" is free and open source software
This project was supported by the Deutsche Forschungsgemeinschaft under Germany’s Excellence Strategy – EXC2068 - Cluster of Excellence "Physics of Life" of TU Dresden. This project has been made possible in part by grant number 2021-240341 (Napari plugin accelerator grant) from the Chan Zuckerberg Initiative DAF, an advised fund of the Silicon Valley Community Foundation.
If you encounter any problems, please file an issue along with a detailed description.
- 07 June 2022
- 15 November 2021
- Stars: 36
- Forks: 5
- Issues + PRs: 15