GraphEcho: Graph-Driven Unsupervised Domain Adaptation for Echocardiogram Video Segmentation

🔨 PostScript

  😄 This project is the pytorch implemention of [paper];

  😆 Our experimental platform is configured with One RTX3090 (cuda>=11.0);

  😊 Currently, this code is avaliable for public dataset CAMUS and EchoNet;

  😃 For codes and accessment that related to dataset CardiacUDA;

      👀 The code is now available at:       ..\datasets\cardiac_uda.py

  😍 For codes and accessment that related to dataset CardiacUDA

      👀 Please follw the link to access our dataset：

💻 Installation

1. You need to build the relevant environment first, please refer to : requirements.yaml

2. Install Environment:

   conda env create -f requirements.yaml
   

• We recommend you to use Anaconda to establish an independent virtual environment, and python > = 3.8.3;

📘 Data Preparation

1. EchoNet & CAMUS

• This project provides the use case of echocardiogram video segmentation task;

• The hyper parameters setting of the dataset can be found in the train.py, where you could do the parameters modification;

• For different tasks, the composition of data sets have significant different, so there is no repetition in this file;

  1.1. Download The CAMUS.

  💬 The detail of CAMUS, please refer to: https://www.creatis.insa-lyon.fr/Challenge/camus/index.html/.

  1. Download & Unzip the dataset.

     The CAMUS dataset is composed as: /testing & /training.

  2. The source code of loading the CAMUS dataset exist in path :

     ..\datasets\camus.py
     and modify the dataset path in
     ..\train_camus_echo.py

     New Version : We have updated the infos.npy in our new released code

  1.2. Download The EchoNet.

  💬 The detail of EchoNet, please refer to: https://echonet.github.io/dynamic/.

  1. Download & Unzip the dataset.

     • The EchoNet dataset is consist of: /Video, FileList.csv & VolumeTracings.csv.

  2. The source code of loading the Echonet dataset exist in path :

     ..\datasets\echo.py
     and modify the dataset path in
     ..\train_camus_echo.py

2. CardiacUDA

1. Please access the dataset through : XiaoweiXu's Github
2. Follw the instruction and download.
3. Finish dataset download and unzip the datasets.
4. Modify your code in both:

   ..\datasets\cardiac_uda.py
   and modify the infos and dataset path in
   ..\train_cardiac_uda.py
   # The layer of the infos dict should be :
   # dict{
   #     center_name: {
   #                  file: {
   #                        views_images: {image_path},
   #                        views_labels: {label_path},}}}

🐾 Training

1. In this framework, after the parameters are configured in the file train_cardiac_uda.py and train_camus_echo.py, you only need to use the command:

   python train_cardiac_uda.py

   And

   python train_camus_echo.py

2. You are also able to start distributed training.

   • Note: Please set the number of graphics cards you need and their id in parameter "enable_GPUs_id".

🚀 Code Reference

• https://github.com/huawei-noah/Efficient-AI-Backbones/tree/master/vig_pytorch
• https://github.com/chengchunhsu/EveryPixelMatters

🚀 Updates Ver 1.0（PyTorch）

🚀 Project Created by Jiewen Yang : jyangcu@connect.ust.hk