Slaid is a library for applying DL models from DeepHealth project (https://deephealth-project.eu/) on WSI.
Prerequisites:
- conda
- python >=3.6, <= 3.8 (tested on 3.8)
- Installation of dependencies pyecvl, pyeddl with conda is recommended. Be sure pip on your path is the one that comes with conda.
Run:
python setup.py install
Run:
make docker
For slide classification, use the installed bin classify.py. Get help typing:
classify.py --help
Examples: Extract tissue
classify.py -f tissue -m slaid/resources/models/tissue_model-extract_tissue_eddl_1.1.bin -l 2 -o <OUTPUT_DIR> <SLIDE>
Classify tumor:
classify.py -f tissue -m slaid/resources/models/tumor_model-classify_tumor_eddl_0.1.bin -l 2 -o <OUTPUT_DIR> <SLIDE>
Slaid is released as docker images, one for each DL model available. Example:
docker run --rm -v $DIR/../data:/data slaid:0.62.0-tissue_model-extract_tissue_eddl_1.1 -l 0 /data/$IMAGE --overwrite -f tissue -o/data
Tests data from https://openslide.cs.cmu.edu/download/openslide-testdata/Mirax/. Credits to Yves Sucaet.