VT-UNet

This repo contains the supported pytorch code and configuration files to reproduce 3D medical image segmentaion results of VT-UNet.

Our previous Code for A Volumetric Transformer for Accurate 3D Tumor Segmentation can be found iside version 1 folder.

VT-UNet: A Robust Volumetric Transformer for Accurate 3D Tumor Segmentation

Parts of codes are borrowed from nn-UNet.

System requirements

This software was originally designed and run on a system running Ubuntu.

Dataset Preparation

• Create a folder under VTUNet as DATASET
• Download MSD BraTS dataset (http://medicaldecathlon.com/) and put it under DATASET/vtunet_raw/vtunet_raw_data
• Rename folder as Task03_tumor
• Move dataset.json file to Task03_tumor

Pre-trained weights

• Download swin-T pretrained weights : https://github.com/SwinTransformer/storage/releases/download/v1.0.8/swin_tiny_patch4_window7_224_22k.pth
• Add it under pretrained_ckpt folder.

Create Environment variables

vi ~/.bashrc

• export vtunet_raw_data_base="/home/VTUNet/DATASET/vtunet_raw/vtunet_raw_data"
• export vtunet_preprocessed="/home/VTUNet/DATASET/vtunet_preprocessed"
• export RESULTS_FOLDER_VTUNET="/home/VTUNet/DATASET/vtunet_trained_models"

source ~/.bashrc

Environment setup

Create a virtual environment

• virtualenv -p /usr/bin/python3.8 venv
• source venv/bin/activate

Install torch

• pip3 install torch==1.10.2+cu113 torchvision==0.11.3+cu113 torchaudio==0.10.2+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html

Install other dependencies

• pip install -r requirements.txt

Preprocess Data

cd VTUNet

pip install -e .

• vtunet_convert_decathlon_task -i /home/VTUNet/DATASET/vtunet_raw/vtunet_raw_data/Task03_tumor
• vtunet_plan_and_preprocess -t 3

Train Model

cd vtunet

• CUDA_VISIBLE_DEVICES=0 nohup vtunet_train 3d_fullres vtunetTrainerV2_vtunet_tumor 3 0 &> small.out &
• CUDA_VISIBLE_DEVICES=0 nohup vtunet_train 3d_fullres vtunetTrainerV2_vtunet_tumor_base 3 0 &> base.out &

Test Model

cd /home/VTUNet/DATASET/vtunet_raw/vtunet_raw_data/vtunet_raw_data/Task003_tumor/

• CUDA_VISIBLE_DEVICES=0 vtunet_predict -i imagesTs -o inferTs/vtunet_tumor -m 3d_fullres -t 3 -f 0 -chk model_best -tr vtunetTrainerV2_vtunet_tumor
• python vtunet/inference_tumor.py vtunet_tumor

Trained model Weights

• VT-UNet-S - (fold 0 only)
• VT-UNet-B (To be updated)

Acknowledgements

This repository makes liberal use of code from open_brats2020, Swin Transformer, Video Swin Transformer, Swin-Unet, nnUNet and nnFormer

References

• Medical segmentation decathlon

Citing VT-UNet

    @misc{peiris2021volumetric,
      title={A Volumetric Transformer for Accurate 3D Tumor Segmentation}, 
      author={Himashi Peiris and Munawar Hayat and Zhaolin Chen and Gary Egan and Mehrtash Harandi},
      year={2021},
      eprint={2111.13300},
      archivePrefix={arXiv},
      primaryClass={eess.IV}
    }