fastMRI: An Open Dataset and Benchmarks for Accelerated MRI
Paper
•
1811.08839
•
Published
•
1
Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 289, in get_dataset_config_info
for split_generator in builder._split_generators(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 64, in _split_generators
with h5py.File(first_file, "r") as h5:
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/h5py/_hl/files.py", line 564, in __init__
fid = make_fid(name, mode, userblock_size, fapl, fcpl, swmr=swmr)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/h5py/_hl/files.py", line 238, in make_fid
fid = h5f.open(name, flags, fapl=fapl)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "h5py/_objects.pyx", line 56, in h5py._objects.with_phil.wrapper
File "h5py/_objects.pyx", line 57, in h5py._objects.with_phil.wrapper
File "h5py/h5f.pyx", line 102, in h5py.h5f.open
FileNotFoundError: [Errno 2] Unable to synchronously open file (unable to open file: name = 'hf://datasets/FLowOak/mri_knee@2bba84ff6436ce36e0466d61c384bd5a4440baa2/train/file1000023.h5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 65, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 343, in get_dataset_split_names
info = get_dataset_config_info(
^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 294, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
🧠 MRI Knee Dataset (Single-Coil) Dataset Name
FLowOak/mri_knee
📌 Dataset Summary
FLowOak/mri_knee is a single-coil knee MRI dataset provided in HDF5 (.h5) format, intended for research in accelerated MRI reconstruction, compressed sensing, and deep learning–based image reconstruction.
This dataset is derived from the fastMRI initiative and adapted for single-coil reconstruction experiments, making it suitable for lightweight models, NAS-based architectures, and academic research where computational resources are limited.
📚 Credits & Acknowledgements
⚠️ Important Credit Notice
This dataset is based on data from the fastMRI dataset, a large-scale open dataset released by Facebook AI Research (FAIR) and NYU Langone Health.
Original Dataset:
fastMRI: An Open Dataset for Accelerated MRI
Authors: Zbontar et al.
Website: https://fastmri.org
License: fastMRI Data License
If you use FLowOak/mri_knee in your research, you must also cite fastMRI in addition to this dataset.
📖 Recommended Citation Cite this dataset: @misc{flowoak_mri_knee, title={MRI Knee Dataset (Single-Coil)}, author={FLowOak}, year={2025}, publisher={Hugging Face}, howpublished={\url{https://huggingface.co/datasets/FLowOak/mri_knee}} }
Also cite fastMRI: @article{zbontar2018fastmri, title={fastMRI: An Open Dataset and Benchmarks for Accelerated MRI}, author={Zbontar, Jure and Knoll, Florian and Sriram, Anuroop and others}, journal={arXiv preprint arXiv:1811.08839}, year={2018} }
🧩 Dataset Structure
File format: .h5 (HDF5)
Imaging modality: MRI (Knee)
Acquisition: Single-coil
Domain: k-space and/or image domain (depending on file)
Resolution: Typically centered and cropped (e.g., 320×320)
ℹ️ The dataset does not currently expose predefined splits (train/validation/test) in the Hugging Face viewer. Users are expected to manage splits manually.
📦 Expected HDF5 Contents
Each .h5 file may contain one or more of the following keys:
Key Description kspace Complex-valued undersampled k-space reconstruction_rss / image Reference image (if available) mask Sampling mask (optional)
⚠️ Users should inspect each file using h5py to confirm exact structure.
🚀 Intended Use
This dataset is intended for research and educational purposes only, including:
Single-coil MRI reconstruction
Compressed sensing MRI
Neural Architecture Search (NAS) for MRI
Image-domain and k-space domain reconstruction
Evaluation using PSNR, SSIM, NMSE
🧪 Example Usage import h5py
with h5py.File("sample.h5", "r") as f: print(list(f.keys())) kspace = f["kspace"][()]
⚙️ Preprocessing Recommendations
Convert complex data into real + imaginary channels
Normalize magnitude values
Apply FFT/IFFT for domain conversion
Generate Cartesian or variable-density masks if not provided
⚠️ Limitations
Single-coil only (not multi-coil clinical reconstruction)
No built-in Hugging Face dataset splits
Requires custom data loading logic
Derived from fastMRI, so original fastMRI license terms apply
🧾 License
This dataset follows the same usage restrictions as fastMRI. Please refer to the original fastMRI license:
👉 https://fastmri.org/licensing/
✅ Final Note
This dataset is not a replacement for fastMRI, but a processed derivative designed to simplify single-coil MRI reconstruction research.
- Downloads last month
- 71