CA_segment

Environment Setup

Create environment and install all dependencies via:

conda create -n cas_env python=3.11.5
conda activate cas_env
cd CA_sgement
pip install -r requirements.txt # (Not Applicable) To-be Updated
pip install -e .

Dataset

We used COCA (Coronary Calcium and chest CT’s) dataset for training and validation. The dataset is available at LINK. We only use gated coronary CT DICOM images with corresponding coronary artery calcium segmentations and scores. (EXCEPT non-gated chest CT DICOM images)

Preprocessing - (Continually Updated)

At first, you run for generating metadata of COCA dataset. This script will generate table including a list of DICOM files and their corresponding metadata and json files including (x,y) coordinates of coronary artery calcium segmentations.

First, you should set up your config file config/config_template.yaml and change it to config/config.yaml. Then, you can run the following command:

python dataset/COCA/preprocess.py --saveFileName <FILENAME>

Dependencies

• python==3.11.5
• torch==2.9.1
• CUDA==12.6
• monai==1.15.1
• wandb==0.21.2
• opencv-python==4.12.0