Skip to content

This is the official code of the paper "Few Shot Medical Image Segmentation with Cross Attention Transformer"

License

Notifications You must be signed in to change notification settings

hust-linyi/CAT-Net

Repository files navigation

Masked Attention Transformer (CAT-Net)

This is the official code for our MICCAI 2023 paper:

Few Shot Medical Image Segmentation with Cross Attention Transformer
Yi Lin*, Yufan Chen*, Kwang-Ting Cheng, Hao Chen

Highlights

In this work, we propose a novel framework for few-shot medical image segmentation, termed CAT-Net, based on cross masked attention Transformer. Our proposed network mines the correlations between the support image and query image, limiting them to focus only on useful foreground information and boosting the representation capacity of both the support prototype and query features. We further design an iterative refinement framework that refines the query image segmentation iteratively and promotes the support feature in turn.

visualization

Left: Overview of the CAT-NET; Right: The architecture of CMAT module.

Using the code

Please clone the following repositories:

git clone https://github.com/hust-linyi/CAT-Net

Requirement

pip install -r requirements.txt

Data preparation

Download

  1. Abdominal CT Synapse Multi-atlas Abdominal Segmentation dataset
  2. Abdominal MRI Combined Healthy Abdominal Organ Segmentation dataset
  3. Cardiac MRI Multi-sequence Cardiac MRI Segmentation dataset (bSSFP fold)

Pre-processing

Please refer to Ouyang et al.

Training

  1. Download pre-trained ResNet-101 weights and put into your own backbone folder.
  2. Run the following command for Abdominal CT/MRI:
sh ./exps/train_Abd.sh

Run the following command for Cardiac MRI:

sh ./exps/train_CMR.sh

Inference

Run ./exp/validation.sh

Visualization

visualization

Qualitative results of our method on Abd-CT and Abd-MRI.

Citation

Please cite the paper if you use the code.

@article{lin2023few,
  title={Few Shot Medical Image Segmentation with Cross Attention Transformer},
  author={Lin, Yi and Chen, Yufan and Cheng, Kwang-Ting and Chen, Hao},
  journal={arXiv preprint arXiv:2303.13867},
  year={2023}}

Acknowledgment

This code is based on Q-Net, PFENet, thanks for their excellent work!

About

This is the official code of the paper "Few Shot Medical Image Segmentation with Cross Attention Transformer"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published