MCDNet: Morphological-Conditional Dual-view Fusion for 3D Tubular Structure Segmentation

Introduction

Approach

Morphological-Conditional Convolution

Dataset

Download the RAOS dataset from here.

Download the CAS2023 dataset from here.

Download the MM-WHS dataset from here.

Download the MSD dataset from here.

Training

Default Scripts

All default hyperparameters among these models are tuned for RAOS datasets.

Wandb is needed if visualization of training parameters is wanted

Customized Execution

run script like this:

python main.py \
--model Our_UNet \
--dataset RAOS \
--batch_size 4 \
--num_epochs 200 \
--learning_rate 1e-4 \
--dropout 0.1 \
--do_train \
--do_evaluate

Dependencies

• python==3.12
• opencv-python==4.7.0.68
• einops
• nilearn==0.10.4
• scikit-learn==1.3.2
• scipy
• torch==2.3.0
• pydicom==2.4.4
• pandas==1.5.3
• nibabel==5.2.1
• wandb

Citation

@ARTICLE{
  author={Wang, Zhiyan and Wang, Changjian and Xu, Kele and Tang, Zhongshun and Zhuang, Yan and Zou, Jiani and Liu, Fangyi},
  journal={}, 
  title={MCDNet: Morphological-Conditional Dual-view Fusion for 3D Tubular Structure Segmentation}, 
  year={2025},
  volume={},
  number={},
  pages={},
  keywords={Tubular Structure Segmentation;Conditional Convolution;Dual-view Architecture},
  doi={}}

Contact Us

If you are interested to leave a message, please feel free to send any email to us at wangzhiyan24@nudt.edu.cn