Deep Learning Identifies Morphological Features in Breast Cancer Predictive of Cancer ERBB2 Status and Trastuzumab Treatment Efficacy
he-erbb2-github-data
βββ tissue-samples
β βββ [ sample 1 ]
β βββ ...
β βββ [ sample n ]
βββ models
βββ Her2
βββ fold-1.pth
βββ fold-2.pth
βββ fold-3.pth
βββ fold-4.pth
βββ fold-5.pth
Attach to the docker container and run:
python inference.py -c configs/inference-config.json
The script generates a .csv
file with predicted scores.
- Clone the Repository
- CD to the docker folder
cd docker
- Edit "DATA_VOLM" path in
init-docker.sh
- point to the data you downloaded through the linkd, e.g./your/path/he-erbb2-github-data/
. Other variables can remain default. - Build an image:
./build-docker.sh
- Run a container:
./run-docker.sh
- Containers run in an interactive mode
- control + p, control + q will turn interactive mode into daemon mode, i.e. detach from a container without stopping it
- Reattach container:
docker attach [name]
Warning:
control + a d β detach from TMUX session when inside a container. This will kill the container!
Scientific Reports PDF
https://www.nature.com/articles/s41598-021-83102-6
DOI: https://doi.org/10.1038/s41598-021-83102-6
Bychkov, D., Linder, N., Tiulpin, A., KΓΌcΓΌkel, H., Lundin, M., Nordling, S., Sihto, H., Isola, J., LehtimΓ€ki, T., KellokumpuβLehtinen, PL., von Smitten, K., Joensuu, H., Lundin, J. Deep learning identifies morphological features in breast cancer predictive of cancer ERBB2 status and trastuzumab treatment efficacy. Sci Rep 11, 4037 (2021)
This code is freely available only for research purposes. Commercial use is not allowed by any means.