CS7643 FastMRI Project

This project works with the fastMRI dataset for MRI reconstruction tasks. The repository contains a Jupyter notebook for exploring and processing the fastMRI knee singlecoil dataset.

Follow these instructions to set up your development environment and get the project running locally.

• Python 3.10
• Conda or Miniconda (recommended for environment management)
• fastMRI dataset (can be downloaded from https://fastmri.med.nyu.edu/)

1. Clone the repository:

   git clone <repository-url>
   cd cs7643-fastmri

2. Create and activate the conda environment using the provided environment.yml file:

   conda env create -f environment.yml
   conda activate cs7643-fastmri

1. Create a copy of the example environment file:

   cp .env.example .env

2. Open the .env file and update the paths to match your local setup:

   SINGLECOIL_TRAIN_PATH=/path/to/your/singlecoil_train
   SINGLECOIL_VAL_PATH=/path/to/your/singlecoil_val
   

1. Start Jupyter Notebook:

   jupyter notebook

2. Open the fastMRI_knee_singlecoil.ipynb notebook

3. Run the cells to explore and process the fastMRI dataset

• don't forget to update .env file (data location)
• when using a new system (with data on the mounted filesystem and not on the local SSD) make sure to copy data to local SSD
• get the environment set up on the mounted filesystem

This assumes that a filesystem (or other directory) is available at ~/fastmri on remote system

rsync -avz --exclude='.git/' --exclude='checkpoints/' --exclude='wandb/' --exclude='__pycache__/' ~/projects/cs7643-fastmri/ ubuntu@<remote-ip>:~/fastmri/code

sudo apt install -y pv pixz
mkdir ~/data
rsync -ah --progress ~/fastmri/data/knee_singlecoil_train.tar.xz ~/fastmri/data/knee_singlecoil_val.tar.xz ~/data/
pixz -d < ~/data/knee_singlecoil_train.tar.xz | tar -x -C ~/data
pixz -d < ~/data/knee_singlecoil_val.tar.xz | tar -x -C ~/data
echo "done"

curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
source ~/.bashrc
conda env create --prefix ~/fastmri/env --file ~/fastmri/code/environment.yml
conda activate ~/fastmri/env

To get a persistent remote session:

tmux new -s fastmri

To attach to a persistent remote session:

tmux attach -t fastmri

To detach from a persistent remote session:

Ctrl+b d

cd ~/fastmri/code
conda activate ~/fastmri/env
python train_model_SCKSCWMT.py