/A-Unified-3D-Cross-Modality-Synthesis-Codebase

This is the codebase for a unified generative methods (CNN-based, GAN-based, and diffusion-based) for 3D medical cross modality synthesis.

Primary LanguagePython

A-Unified-3D-Cross-Modality-Synthesis-Codebase

This is the codebase for a unified generative methods (CNN-based, GAN-based, and diffusion-based) for 3D medical cross modality synthesis.

If you find our project useful, please 🌟 our projects and cite the following paper:

@article{wang2024joint,
  title = {Joint learning framework of cross-modal synthesis and diagnosis for Alzheimer’s disease by mining underlying shared modality information},
  author = {Wang, Chenhui and Piao, Sirong and Huang, Zhizhong and Gao, Qi and Zhang, Junping and Li, Yuxin and Shan, Hongming and others},
  journal = {Medical Image Analysis},
  volume = {91},
  pages = {103032},
  year = {2024},
  publisher = {Elsevier}
}

Todos

  • Publish the multi-threads preprocessing code for 3D MRI and PET brain images.
  • Publish CNN-based 3D medical cross-modality synthesis codes. (UNet, DenseUNet, SwinUNetr, etc.)
  • Publish GAN-based 3D medical cross-modality syhtesis codes. (Pix2Pix, CycleGAN, ErGAN, ShareGAN, MultiShareGAN, etc).
  • Publish Sora/DiT-version Diffusion-based 3D medical cross-modality synthesis codes.
  • Publish 3D evaluation methods (MAE, PSNR, SSIM).
  • Publish a series of brain analysis and interpretable methods.

Updates

  • April, 2024: Initial commit.
  • May, 2024: update data preprocessing codes for 3D MRI and PET brain images.
  • May, 2024: update basic 3D evaluation methods.