Carotid Artery Segmentation using Center point Annotation

Overview

Dataset

We used CosMos 2022 Challenge Dataset

please download the dataset through this

Train dataset: https://zenodo.org/record/6481870#.Y7aqaXZBzdk

Test dataset: https://zenodo.org/record/6843257#.Y7aqSHZBzdk

How to use

1. Download the data set from the link above and preprocess the dataset

2. Extract the center point from the given label

3. Make a learning plan using Totalloss at loss.py.

Train

python main.py --mode total --model Unet3D --fold 0 

Pretrained Model

model weight: google drive

Citation

@article{kim2025weakly,
  title={Weakly-supervised segmentation using sparse single point annotations for lumen and wall of carotid arteries in 3D MRI},
  author={Kim, Jonghun and Na, Inye and Kwon, Junmo and Seo, Woo-Keun and Park, Hyunjin},
  journal={Computer Methods and Programs in Biomedicine},
  pages={108881},
  year={2025},
  publisher={Elsevier}
}