FlowVM-Net

This is the official code repository for "FlowVM-Net: Enhanced Vessel Segmentation in X-Ray Coronary Angiography Using Temporal Information Fusion", which is accpeted by Journal of Imaging Informatics in Medicine as a original paper!

Framework Overview

Figure 1: The overall architecture of FlowVM-Net combining spatial features with temporal information

0. Environments.

conda create -n flowvmnet python=3.8
conda activate flowvmnet
pip install torch==1.13.0 torchvision==0.14.0 torchaudio==0.13.0 
pip install packaging
pip install timm==0.4.12
pip install pytest chardet yacs termcolor
pip install submitit tensorboardX
pip install triton==2.0.0
pip install causal_conv1d==1.0.0  # causal_conv1d-1.0.0+cu118torch1.13cxx11abiFALSE-cp38-cp38-linux_x86_64.whl
pip install mamba_ssm==1.0.1  # mmamba_ssm-1.0.1+cu118torch1.13cxx11abiFALSE-cp38-cp38-linux_x86_64.whl
pip install scikit-learn matplotlib thop h5py SimpleITK scikit-image medpy yacs opencv-python

The .whl files of mamba_ssm could be found here. The .whl files of causal_conv1d could be found here.

1. Prepare the dataset.

Data Format

├── './data/your_dataset/'
    ├── train
        ├── images
            ├── 001.png
            ├── 002.png
            └── ... # {id}.png 
        ├── masks
            ├── 001.png
            ├── 002.png
            └── ... # {id}.png 
    ├── val
        ├── images
            ├── 001.png
            ├── 002.png
            └── ... # {id}.png 
        ├── masks
            ├── 001.png
            ├── 002.png
            └── ... # {id}.png 
    ├── test
        ├── images
            ├── 001.png
            ├── 002.png
            └── ... # {id}.png 
        ├── masks
            ├── 001.png
            ├── 002.png
            └── ... # {id}.png 

2. Train the FlowVM-Net.

• The weights of the pre-trained VMamba could be downloaded here.
• The weights of the pre-trained Optical Flow model could be downloaded here. After that, the pre-trained weights should be stored in './pre_trained_weights/'.

bash train.sh

• After trianing, you could obtain the outputs in ./results/

3. Test the FlowVM-Net.

First, in the testing.py file, you should change the address of the checkpoint in 'checkpoint path'.

python testing.py

4. Citation

Please cite the paper as follows if you use the code from FlowVM-Net:

@article{wei2025flowvm,
  title={FlowVM-Net: Enhanced Vessel Segmentation in X-Ray Coronary Angiography Using Temporal Information Fusion},
  author={Guangyu Wei and Xueying Zeng and Qing Zhang},
  journal={Journal of Imaging Informatics in Medicine},
  year={2025},
  publisher={Springer},
  doi={10.1007/s10278-025-01732-y}
}