MCTrans

Multi-Compound Transformer for Accurate Biomedical Image Segmentation

Introduction

  • This repository provides code for "Multi-Compound Transformer for Accurate Biomedical Image Segmentation" [paper].

  • The MCTrans repository heavily references and uses the packages of MMSegmentation, MMCV, and MONAI. We thank them for their selfless contributions

Highlights

  • A comprehensive toolbox for medical image segmentation, including flexible data loading, processing, modular network construction, and more.

  • Supports representative and popular medical image segmentation methods, e.g. UNet, UNet++, CENet, AttentionUNet, etc.

Changelog

The first version was released on 2021.7.16.

Model Zoo

Supported backbones:

  • [x] VGG
  • [x] ResNet

Supported methods:

  • [x] UNet
  • [x] UNet++
  • [x] AttentionUNet
  • [x] CENet
  • [x] TransUNet
  • [x] NonLocalUNet

Installation and Usage

Please see the guidance.md.

Citation

If you find this project useful in your research, please consider cite:

@article{ji2021multi,
  title={Multi-Compound Transformer for Accurate Biomedical Image Segmentation},
  author={Ji, Yuanfeng and Zhang, Ruimao and Wang, Huijie and Li, Zhen and Wu, Lingyun and Zhang, Shaoting and Luo, Ping},
  journal={arXiv preprint arXiv:2106.14385},
  year={2021}
}

GitHub

https://github.com/JiYuanFeng/MCTrans