EBV-TRACER

Environment Setup

conda create -n ebv_tracer python=3.9
conda activate ebv_tracer
pip install -r requirements.txt

Patch Generation

python make_patches.py \
--input_svs_dir INPUT_SVS_DIR \
--save_patch_dir SAVE_PATCH_DIR \
--n_jobs 16

Two stage stain normalization

python two_stage_stain_normalize.py \
--target_svs_dir TARGET_SLIDE_DIR \
--input_svs_dir SOURCE_SLIDE_DIR \
--input_patch_dir SOURCE_PATCH_DIR \
--output_patch_dir OUTPUT_PATCH_DIR \
--output_thumbnail_dir OUTPUT_THUMBNAIL_DIR \
--n_jobs 16

EBV-GC Classification

Download pre-trained EfficientNet B1 model weights from this link and save ot in the "./checkpoints" directory.

Run Inference

python inference.py \
--workers 16 \
--batch_size 128 \
--checkpoint ./checkpoints/net_5000.pth \
--result_path ./results.csv \
--data PATCH_DIR