The spherical-patches-extraction and deep-learning based neuron reconstructor (SPE-DNR) is an automatic method for neuron reconstruction from 3D microscopy images.

This is the source code for the SPE-DNR with pretrained weights. And here is the full supplementary material of the paper.

• pytorch==1.7.1
• numpy==1.19.2
• scipy==1.5.2
• scikit-image==0.17.2
• libtiff
• tqdm

python main.py

In folder './checkpoint/classification_checkpoints'.

One test image with corresponding seed maps and soma masks is in folder './test_samples'.

If the code or method help with your research, please cite the following paper:

@article{Chen2021deep,
  title={Deep-learning based automated neuron reconstruction from 3D microscopy images using synthetic training images},
  author={Chen, Weixun and Liu, Min and Du, Hao and Radojević, Miroslav and Wang, Yaonan and Meijering, Erik},
  journal={IEEE Transactions on Medical Imaging},
  year={2021},
  doi={https://doi.org/10.1109/TMI.2021.3130934}
}