MSF-WSI: A multi-resolution self-supervised learning framework for semantic segmentation in histopathology

Officail repo for MSF-WSI.

Note

Current repo provides short documentation and will provide more details when I have time... Feel free to email me for any supports

Installation

Run the following code to install all dependencies.

conda env create -f environment.yml

Steps

Check scripts in /script to see how to run MSF-WSI pre-training for BCSS, PAIP19, and Camelyon16 datasets.

Prepare BCSS dataset and generate csv file by

python -u ./tools/bcss_prepare.py -p RAW_DATA_PATH -o ../data/bcss/L0_1024_s512 -s 1024 --overlap 512

Model Weights

Dataset	SSL pre-trained	Fine-tuned
BCSS (fold 1)	Link	Link
PAIP2019 (fold 1)	Link	Link
Camelyon16	Link	N/A