The espm package is designed for simulation and physics-guided NMF decomposition of hyperspectral data. Even though the package is mainly centered around electron spectro-microscopy applications, custom models can be implemented for other type of data. Currently espm supports the simulation and analysis of simultaneous scanning transmission electron microscopy and energy dispersive X-ray spectroscopy (STEM / EDXS). In future implementation, we will try to extend the package to support electron energy loss spectroscopy (EELS).
This library is integrated as much as possible in the hyperspy <https://hyperspy.org> and scikit-learn <https://scikit-learn.org> frameworks.
The main components of the package are: - The simulation of STEM-EDXS datasets using espm.datasets which combines espm.weights for the simulation of spatial distributions and espm.models for the simulation of spectra. - The hyperspectral unmixing of STEM-EDXS spectrum images using espm.estimators. This module contains algorithms to perform non-negative matrix factorization with diverse regularisation (e.g. Laplacian or L1) and contraints (e.g. simplex). - The espm.models module can also be used to perform a physics-guided decomposition of STEM-EDXS datasets.
You can install this package from PyPi using:
$ pip install espm
If you want to develop, please use the option:
$ git clone https://github.com/adriente/espm.git $ cd espm $ pip install cython $ pip install -e .[dev]
If you get issues regarding pandoc when using make doc, you can install it using:
$ sudo apt-get install pandoc
or:
$ conda install pandoc
We recommend to install the package in a virtual environment using conda:
$ conda create -n espm python=3.11 $ conda activate espm $ pip install espm $ conda install jupyterlab
It is especially useful for the interactive plotting in the notebooks.
Try the api.ipynb notebook in the notebooks folder.
The documentation is available at https://espm.readthedocs.io/en/latest/
You can get started with the following notebooks:
- Simulate STEM-EDXS data : https://espm.readthedocs.io/en/latest/introduction/notebooks/generate_data.ipynb
- Physics-guided decomposition (ESpM-NMF) STEM-EDXS data : https://espm.readthedocs.io/en/latest/introduction/notebooks/api.html
- Tests of the ESpM-NMF with a toy dataset : https://espm.readthedocs.io/en/latest/introduction/notebooks/toy-problem.html
If you use this library, please cite on of the following papers:
@article{teurtrie2023espm,
title={espm: A Python library for the simulation of STEM-EDXS datasets},
author={Teurtrie, Adrien and Perraudin, Nathana{\"e}l and Holvoet, Thomas and Chen, Hui and Alexander, Duncan TL and Obozinski, Guillaume and H{\'e}bert, C{\'e}cile},
journal={Ultramicroscopy},
pages={113719},
year={2023},
publisher={Elsevier}
}
@article{Teurtrie_2024,
doi = {10.1088/2632-2153/ad9192},
year = {2024},
month = {nov},
publisher = {IOP Publishing},
volume = {5},
number = {4},
pages = {045050},
author = {Teurtrie, Adrien and Perraudin, Nathanaël and Holvoet, Thomas and Chen, Hui and Alexander, Duncan T L and Obozinski, Guillaume and Hébert, Cécile},
title = {From STEM-EDXS data to phase separation and quantification using physics-guided NMF},
journal = {Machine Learning: Science and Technology}
}