[Under Reorganization] The code is not fully cleaned and organized. We will clean it up and release a more detailed readme file.
SIB-MIL: Sparsity-Induced Bayesian Neural Network for Robust Multiple Instance Learning on Whole Slide Image Analysis
We use three dataset projects in our paper for demonstration: 1) Camelyon16, 2) TCGA and 3) BRACS.
You may follow the instructions in the websites to download the data.
We crop slides with magnification parameter set to 20 (level 0) and features are extracted using pretrained KimiaNet. We followed the pipeline of DSMIL.
python main.py --backbone abmil --num_epochs 100 --dataset BRCA --task staging --feats_size 1024 --extractor Kimia --num_workers 1 --num_rep 1 --wandb