Two-Stage Diffusion Model for 3D Medical Image Segmentation

This model is a novel two-stage diffusion-based segmentation model that retains its original structure with reduced computational costs. And, The paper of this has been officially accepted by IJCNN 2024.

Model overview

Requirement

Python >= 3.8.16

pip install -r requirement.txt

How to use

1. Go to BraTS official cite and download the brain MRI data.
2. Rewrite args.py as needed for information such as data storage location.
3. Run the following:

  python two_stage_train1st.py
  python two_stage_createData1st.py
  python two_stage_train2nd.py
  python two_stage_createData2nd.py

Evaluation

Evaluate according to this repository, which is the official performance metrics program.