lukmuk/r_em_workflows

Example workflows for using the r_em neural network for image denoising in electron microscopy

Denoise/Restore Electron-Microscopy Images using the r_em Deep-Learning Model by Lobato et al.

This notebook contains ideas for denoising workflows using the model and guidelines from the official repository (https://github.com/Ivanlh20/r_em).
It shows how to use tifffile to load TIFF images, denoise/restore them, and save them back to TIFF.
There is also an example showing how to use RosettaSciIO to directly read .emi/.ser or .dm3/.dm4 files to numpy array for subsequent denoising.

Paper: https://www.nature.com/articles/s41524-023-01188-0

Please cite their work in your publications if it helps your research:

@article{Lobato2024,
   author = {I. Lobato and T. Friedrich and S. Van Aert},
   doi = {10.1038/s41524-023-01188-0},
   issn = {2057-3960},
   issue = {1},
   journal = {npj Computational Materials 2024 10:1},
   keywords = {Imaging techniques,Transmission electron microscopy},
   month = {1},
   pages = {1-19},
   publisher = {Nature Publishing Group},
   title = {Deep convolutional neural networks to restore single-shot electron microscopy images},
   volume = {10},
   url = {https://www.nature.com/articles/s41524-023-01188-0},
   year = {2024},
}

Example of SrTiO$_3$[100] (HAADF, FEI Tecnai Osiris, 200 keV):