ce-coronary-artery

Repository for automatic segmentation of contrast enhanced (ce) coronary arteries.

Overview

• This projet utilizes the ImageCAS dataset
  • Contains 1000 contrast enhanced CT angiography images with 1000 corresponding coronary artery labels
• The Hausdorff loss function is utilized throughout the training process
• The (.....TBD) model is utilized

Goals

• Introduce a state of the art coronary artery segmentation model
  • Measured in terms of dice coefficient, hausdorff distance, overlap, ...
• Introduce the vendor neutral, GPU optimizzed Hausdorff loss function

Environment Setup

Prerequisite

You should login in to the cluster.
You can be in any node(server). I recommand you to login to the free CPU server.

Install Mini-conda

We are using Mini-conda for enviornment.

• run cd first to get to the default path.
• then run the following commands.
• Before running: if you don't want to install conda to directory /miniconda, change the path in the scripts.

mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh

After running the scripts:

• run ~/miniconda3/bin/conda init bash

Now, you should be able to use conda commands.

• run conda --version to make sure your conda is ready.
• if you are still not able to use conda commands, try source ~/.bashrc, which restart/refresh the bash shell.

Create Env

• conda create --name imagecas python=3.9 to create a virtual environment with python version 3.9
• conda activate imagecas to run the virtual environment
• conda install pip to install pip package manager for the virtual environment
• you have to check if you are using the correct pip by pip --version and see the path of the pip. It should be some path that contains miniconda3 like .../miniconda3/env/imagecas...
• if your pip is not having the correct path, contact Liam. (I think you should delete the pyenv settings in .bashrc or .bash_profile Andrew :))
• [unnecessary] you can check if you are using the correct python version by python --version and see the path of the pip

Install Dependencies

• cd to the directory you clone this repo. If you didn't clone the repo, clone it first, then open it in terminal.
• run pip install -r requirements.txt
• done :)

Run the project with marimo

• in your environment, do pip install marimo
• run marimo edit [filename] --host 0.0.0.0 --port [num] to run the marimo file, an example is marimo edit newTest.py --host 0.0.0.0 --port 1234
• do ssh port forwarding mentioned in Zulip group chat