Overview

The proposed script gets ground truth images in hdf5 format from the fastMRI dataset and unpack them to target folder.

The fastMRI dataset is a large open dataset with MRI of the knees and brains.

Learn more about fastMRI project and how to get fastMRI dataset - https://fastmri.org/.

Requirements

• curl - $ sudo apt install curl
• Python 3.5 or latter
• watchdog - $ pip install watchdog
• h5py - $ pip install h5py

Download

$ git clone https://github.com/lazyniv/fastMRI_gt_parser.git

Usage

$ cd fast_MRI_gt_parser/
$ sh run.sh "<URL/to/tar/archive>" <path/to/target/folder>

Description of downloaded data

Let old_file.h5 it's a file before downloading.

After downloading the file to the target directory is named old_file_gt.h5 and this file contains one key - gt_tensor(key according to hdf5 format).