This code has 2 versions:

🦒 Colab version (Recommend): inference and training mode! Fast and easy to run

👨‍💻 GitHub version: inference mode, but need to set up the environment.

📖 🦒 🗂️ 🕹️

VASeg: vessel-aneurysm-segmentation

Cerebrovascular and Aneurysm Segmentation in 3DRA images via a Deep Multi-Task Network

Weights can be download here:

fmnet5.hdf5 for inference_AneuristNet.py

fmnet84.hdf5 for inference_Transformer.py

Folder structure:

vessel-aneurysm-segmentation/
├── data
│   └── step0 (put the data you want to do inference here)
│   │   └── ANSYS_UNIGE_09_image.nii.gz
│   │   ├── ANSYS_UNIGE_28_image.nii.gz
│   │   └── ...
│   ├── step1 (pre-processed data, generate automatically)
│   │   └── ANSYS_UNIGE_09_image.nii.gz
│   │   ├── ANSYS_UNIGE_28_image.nii.gz
│   │   └── ...
├── inference
│   └── inference.ipynb (use this to do inference)
│   ├── inference.py
├── results
│   └── aneurysm (aneurysm prediction)
│   │   └── ANSYS_UNIGE_09_image-[360, 633].nii.gz
│   │   ├── ANSYS_UNIGE_28_image-[1540].nii.gz
│   │   └── ...
│   ├── vessel (vessel prediction)
│   │   └── ANSYS_UNIGE_09_image_vessel_59969.nii.gz
│   │   ├── ANSYS_UNIGE_28_image_vessel_68437.nii.gz
│   │   └── ...
├── weights
│   └── fmnet5.hdf5
└── requirements.txt

How to use the code:

cd vessel-aneurysm-segmentation
pip install -r requirements.txt
cd inference
python inference_AneuristNet.py

Citation ヾ(o′▽`o)ノ°°

@article{lin2023high,
  title={High-throughput 3DRA segmentation of brain vasculature and aneurysms using deep learning},
  author={Lin, Fengming and Xia, Yan and Song, Shuang and Ravikumar, Nishant and Frangi, Alejandro F},
  journal={Computer Methods and Programs in Biomedicine},
  volume={230},
  pages={107355},
  year={2023},
  publisher={Elsevier}
}