X-TRON404 / MSAI-practicum

MSAI-Practicum project repository.

pneumonitis_prediction

Environment setup

• Step. 1

  Visit the link https://services.northwestern.edu/TDClient/30/Portal/KB/ArticleDet?ID=1112, and follow the instructions under Pytorch in Anaconda of What GPU software is available on QUEST? to create a conda environment with Pytorch installed.

• Step. 2

  conda activate pytorch-1.11-py38
  pip install -r ./requirements.txt
  

• Step. 3

  replace the environment path of the two .sh files in the code folder (run-info.sh, run-train.sh) with your own path

Files Location

• code (demo is not on server): /home/phv0465/workspace/project-deliverable/code
• data: lung (/projects/p32050/lung); dose (/projects/p32050/dose); label (/projects/p32050/all_stats.csv)

Train Model

On Northwestern Quest server

cd code
sbatch ./run-train.sh

FPS/#Params/MACs Measurement

On Northwestern Quest server

cd code
sbatch ./run-info.sh

Notebook

Please use notebook.ipynb in code folder for prediction and visualization.

Demo

On a server with GPU in Northwestern. Have access to medical data.

cd demo
streamlit run ./demo.py

access the link shown in terminal for web interface.