EMcapsulins segmentation

Deep learning models to segment emcapsulins in 2D TEM micrograph images from the manuscript:

Genetically encoded barcodes for correlative volume electron microscopy

Expected data

The models are trained on 8-bit images with a pixel size: of 0.5525 per nanometer. We include some example input and output files for replication.

Installation

1. Clone this repository:

   git clone https://github.com/HelmholtzAI-Consultants-Munich/EMcapsulins_segmentation.git

2. Go into the repository and install:

   cd EMcapsulins_segmentation
   pip install -r requirements.txt
   pip install -e .
   

Usuage

run_inference_tiff.py <-- Example script for inference creating tiff files

run_inference_nifti.py <-- Example script for inference creating nifti files

Citation

when using the software please cite https://www.nature.com/articles/s41587-023-01713-y

@article{sigmund2023genetically,
  title={Genetically encoded barcodes for correlative volume electron microscopy},
  author={Sigmund, Felix and Berezin, Oleksandr and Beliakova, Sofia and Magerl, Bernhard and Drawitsch, Martin and Piovesan, Alberto and Gon{\c{c}}alves, Filipa and Bodea, Silviu-Vasile and Winkler, Stefanie and Bousraou, Zoe and others},
  journal={Nature Biotechnology},
  pages={1--12},
  year={2023},
  publisher={Nature Publishing Group US New York}
}

Recommended Environment

• CUDA 11.4+
• Python 3.10+
• GPU with at least 8GB of VRAM

further details in requirements.txt

train your own EMcapsulin segmentation network

Please have a look at this repository

Licensing

This project is licensed under the terms of the GNU Affero General Public License v3.0.

Contact us regarding licensing.

Contact / Feedback / Questions

If possible please open a GitHub issue here.

For inquiries not suitable for GitHub issues:

felix.sigmund @ tum .de

gil.westmeyer @ tum .de