MHIM-MIL [Updating]

Official repo of Multiple Instance Learning Framework with Masked Hard Instance Mining for Whole Slide Image Classification, ICCV 2023. [arXiv] [Slide] [Oral]

News

I've updated the preprocess parameters of TCGA datasets. These parameters perform better on the NSCLC sub-project.
I've found that the Baidu Cloud dataset link often fail, if you've found this you can leave an issue and I'll keep it updated. You can also check this link to download the data you need.
Another work, Feature Re-Embedding: Towards Foundation Model-Level Performance in Computational Pathology, about computational pathology has been accepted by CVPR2024. [arXiv] [Code]
Updating DOCKERFILE (Docker image has been uploaded!)
Uploaded datasets, only including features
Uploaded paper slides and video

Prepare Patch Features

To preprocess WSIs, we used CLAM.

Download the preprocessed patch features: Baidu Cloud, Backup Link.

Patching

--preset bwh_biopsy.csv for C16 (It's the preset parameters officially provided by CLAM), --preset preprocess_tcga_nsclc.csv for TCGA-NSCLS (It's the customized parameters).

# for C16
python create_patches_fp.py --source DATA_DIRECTORY --save_dir RESULTS_DIRECTORY --patch_size 512 \
--step_size 512 --preset bwh_biopsy.csv --seg --patch
# for TCGA-NSCLC
python create_patches_fp.py --source DATA_DIRECTORY --save_dir RESULTS_DIRECTORY --patch_size 512 \
--step_size 512 --preset preprocess_tcga_nsclc.csv --seg --patch

Feature Extraction

CUDA_VISIBLE_DEVICES=0,1 python extract_features_fp.py --data_h5_dir DIR_TO_COORDS --data_slide_dir \
DATA_DIRECTORY --csv_path CSV_FILE_NAME --feat_dir FEATURES_DIRECTORY --batch_size 512 --target_patch_size=224 --slide_ext .svs

Training

Download the Docker Image: Baidu Cloud.

Prepare Your Own Initiation Weight

model=pure baseline=[attn,selfattn,dsmil]

# transmil baseline on C16 dataset
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=3 --title=transmil --model=pure --baseline=selfattn --seed=2021
# transmil baseline on TCGA-NSCLC dataset
python3 main.py --project=$PROJECT_NAME --datasets=tcga --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=4 --val_ratio=0.13 --title=transmil --model=pure --baseline=selfattn --seed=2021
# abmil baseline on C16 dataset
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=3 --title=abmil --model=pure --baseline=attn --seed=2021
# DSMIL baseline on C16 dataset
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=3 --title=dsmil --model=pure --baseline=dsmil --seed=2021

Training MHIM-MIL

--cv_fold=3 --val_ratio=0 for C16, --cv_fold=4 --val_ratio=0.13 for TCGA-NSCLC

ABMIL

# C16
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=3 \
--teacher_init=./modules/init_ckp/c16_3fold_init_abmil_seed2021 --title=abmil_101_mr50h1-0r50_is --baseline=attn \
--num_workers=0 --cl_alpha=0.1 --mask_ratio_h=0.01 --mask_ratio_hr=0.5 --mrh_sche --init_stu_type=fc --mask_ratio=0.5 --mask_ratio_l=0. --seed=2021
# TCGA-NSCLC
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=4 \
--val_ratio=0.13 --teacher_init=./modules/init_ckp/tcga_4fold_init_abmil_seed2021 --title=abmil_105_mr70l20h2-0_mmcos_is \
--baseline=attn --num_workers=0 --cl_alpha=0.5 --mask_ratio_h=0.02 --mrh_sche --mm_sche --init_stu_type=fc --mask_ratio=0.7 --mask_ratio_l=0.2 --seed=2021

TransMIL

# C16
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=3 \
--teacher_init=./modules/init_ckp/c16_3fold_init_transmil_seed2021 --mask_ratio_h=0.03 --mask_ratio_hr=0.5 \
--mrh_sche --title=transmil_101_sml80h3-0r50_mmcos_is --mask_ratio=0. --mask_ratio_l=0.8 --cl_alpha=0.1 --mm_sche --init_stu_type=fc --attn_layer=0 --seed=2021
# TCGA-NSCLC
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=4 \
--val_ratio=0.13 --teacher_init=./modules/init_ckp/tcga_4fold_init_transmil_seed2021 --mask_ratio_h=0.02 \
--mrh_sche --title=transmil_105_smr60l20h2-0_is --mask_ratio=0.6 --mask_ratio_l=0.2 --cl_alpha=0.5 --init_stu_type=fc --attn_layer=0 --seed=2021

DSMIL (same hyper-param with ABMIL, maybe not optimal)

# C16
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=3 \
--teacher_init=./modules/init_ckp/c16_3fold_init_dsmil_seed2021 --title=dsmil_101_mr50h1-0r50_is --baseline=dsmil \
--num_workers=0 --cl_alpha=0.1 --mask_ratio_h=0.01 --mask_ratio_hr=0.5 --mrh_sche --init_stu_type=fc --mask_ratio=0.5 --mask_ratio_l=0. --seed=2021
# TCGA-NSCLC
python3 main.py --project=$PROJECT_NAME --dataset_root=$DATASET_PATH --model_path=$OUTPUT_PATH --cv_fold=4 \
--val_ratio=0.13 --teacher_init=./modules/init_ckp/tcga_4fold_init_dsmil_seed2021 --title=dsmil_105_mr70l20h2-0_mmcos_is \
--baseline=dsmil --num_workers=0 --cl_alpha=0.5 --mask_ratio_h=0.02 --mrh_sche --mm_sche --init_stu_type=fc --mask_ratio=0.7 --mask_ratio_l=0.2 --seed=2021

Citing MHIM-MIL

@InProceedings{Tang_2023_ICCV,
    author    = {Tang, Wenhao and Huang, Sheng and Zhang, Xiaoxian and Zhou, Fengtao and Zhang, Yi and Liu, Bo},
    title     = {Multiple Instance Learning Framework with Masked Hard Instance Mining for Whole Slide Image Classification},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2023},
    pages     = {4078-4087}
}

DearCaat / MHIM-MIL

readme