STATSTEM is an open source user friendly software (Github repository) to quantify scanning transmission electron microscopy (STEM) images by using model based fitting. This provides an accurate and precise way of extracting quantitative information from STEM images. The software is described in this publication: 10.1016/j.ultramic.2016.08.018
– The maximum a posteriori (MAP) probability rule, which has been proposed as an objective, and the quantitative method to detect atom columns and even single atoms from STEM images, have been implemented as an extra plugin in the freely available StatSTEM software. More details can be found in the corresponding publications 10.1103/PhysRevLett.121.056101 and 10.1016/j.ultramic.2019.02.003.
– The hidden Markov model, which can be used to measure atomic scale dynamics of nanoparticles from experimental high-resolution annular dark field STEM images, has been implemented as an extra plugin in the freely available StatSTEM software. More details can be found in the corresponding publications: 10.1016/j.ultramic.2020.113131 and 10.1103/PhysRevLett.124.106105.
MULTEM provides a collection of routines written in C++ with CUDA (Github repository) to perform accurate and fast multislice simulations for different TEM experiments as: HRTEM, STEM, ISTEM, ED, PED, CBED, ADF-TEM, ABF-HC, EFTEM and EELS. The software is developed by Dr. Ivan Lobato and can be found on Github. The software is described these publications: 10.1016/j.ultramic.2015.04.016, 10.1016/j.ultramic.2016.06.003
The Github repository of Thomas Friedrich contains development level contributions to software relevant to the project: https://github.com/ThFriedrich
TK_R_EM: A deep convolutional neural network has been trained to correct for scan distortions. This machine learning approach shows impressive restoration results even for combinations of high levels of distortions for both, periodic and non-periodic structures. The trained models as well as the script used for training are openly accessible. The repository can be accessed at the following URL: https://github.com/Ivanlh20/tk_r_em
More details can be found in the following publication: https://www.nature.com/articles/s41524-023-01188-0
RT_PPISCS: Development of a CNN to predict scattering cross sections, which are often used to quantify STEM images in terms of composition or thickness by matching experimental scattering cross sections with theoretically simulated ones. The inference code for MATLAB, python and the tensorflow source code for training is available at the following URL:
https://github.com/Ivanlh20/RT_PPISCS
More details can be found in the following publication: https://doi.org/10.1016/J.ULTRAMIC.2023.113769
RiCOM: Development of two highly dose-efficient real-time imaging methods in 4D STEM: riCOM algorithm for fast and accurate reconstruction of the projected atomic potential and the AIRPI method using CNNs.
The source code and datasets used in the riCOM study are available at the following URLs:
https://github.com/ThFriedrich/riCOM_cpp
https://doi.org/10.5281/zenodo.5572123
More details can be found in the following publication: https://doi.org/10.1017/S1431927622000617
The AIRPI trained neural network, reconstruction implementations, training datasets, and data generation code are publicly available at the following URLs: https://github.com/ThFriedrich/airpi
https://doi.org/10.5281/zenodo.6971200
https://github.com/ThFriedrich/ap_data_generation
More details can be found in the following publication: https://doi.org/10.1093/MICMIC/OZAC002