tcga_stroma_prediction

Steps to create masks from WSIs

1. Clone repo.

• git clone https://github.com/hwanglab/tcga_stroma_prediction.git

2. Move to "tcga_stroma_prediction" directory

3. Run python

• python normalization.py 'input directory' 'output file name' 'tile size' ** Note that this will create a .csv file with mu and std statistics to be used as an input.
• python generate_mask_from_WSIs.py 'input directory' 'output directory' 'model path' 'norm stats' ** Use the .csv file for 'norm stats'

4. Confirm the masks in 'your output directory'

** Note that you may need to install some python packages.

check python version and location

python --version

create virtual environment

virtualenv -p /usr/bin/python3.8 env

activate

source env/bin/activate

package versions

numpy==1.19.5 tensorflow-gpu==2.6.2 scipy==1.4.1 scikit-image==0.17.2 scikit-learn==0.24.2 pandas==1.1.5 Pillow==8.4.0