MURA Dataset

Estimated reading: 3 minutes

Visualization of the MURA dataset in the Deep Lake UI

MURA dataset

What is MURA Dataset?

MURA is a large musculoskeletal radiograph dataset containing 40,561 images from 14,863 exams, where each exam is labeled normal or abnormal by radiologists themselves. Researchers can use this dataset to see if their models perform as accurately as radiologists on the task.

Download MURA Dataset in Python

Instead of downloading the MURA dataset in Python, you can effortlessly load it in Python via our Deep Lake open-source with just one line of code.

Load MURA Dataset Training Subset in Python

				
					import deeplake
ds = deeplake.load("hub://activeloop/mura-train")

Load MURA Dataset Validation Subset in Python

				
					import deeplake
ds = deeplake.load("hub://activeloop/mura-val")

MURA Dataset Structure

MURA Data Fields

images: tensor containing the image.
region: tensor containing the name of one of the seven regions of studies such as elbow, finger, forearm, hand, humerus, shoulder, and wrist to which the image belongs to.
study: tensor containing the studies.
study_type: tensor containing the study type which can be either positive or negative.
patient_id: tensor containing the id of the patient.

MURA Data Splits

The MURA dataset training set is composed of 11,184 patients, 13,457 studies, and 36,808 images.
The MURA dataset validation set is composed of 783 patients, 1,199 studies, and 3,197 images.

How to use MURA Dataset with PyTorch and TensorFlow in Python

Train a model on MURA dataset with PyTorch in Python

Let’s use Deep Lake built-in PyTorch one-line dataloader to connect the data to the compute:

				
					dataloader = ds.pytorch(num_workers=0, batch_size=4, shuffle=False)

Train a model on MURA dataset with TensorFlow in Python

				
					dataloader = ds.tensorflow()

MURA Dataset Creation

Data Collection and Normalization Information

To evaluate models and assess radiologist performance, additional labels from six certified Stanford radiologists were collected in a test set of 207 musculoskeletal exams. In this test set, the gold standard is a majority vote of a group of three radiologists.

Additional Information about MURA Dataset

MURA Dataset Description

Homepage: https://stanfordmlgroup.github.io/competitions/mura/
Repository: N/A
Paper: MURA: Large Dataset for Abnormality Detection in Musculoskeletal Radiographs, Pranav Rajpurkar*, Jeremy Irvin*, Aarti Bagul, Daisy Ding, Tony Duan, Hershel Mehta, Brandon Yang, Kaylie Zhu, Dillon Laird, Robyn L. Ball, Curtis Langlotz, Katie Shpanskaya, Matthew P. Lungren, Andrew Y. Ng
Point of Contact: ht tps://groups.google.com/forum/#!forum/mura-dataset

MURA Dataset Curators

Pranav Rajpurkar, Jeremy Irvin, Aarti Bagul, Daisy Ding, Tony Duan, Hershel Mehta, Brandon Yang, Kaylie Zhu, Dillon Laird, Robyn L. Ball, Curtis Langlotz, Katie Shpanskaya, Matthew P. Lungren, Andrew Y. Ng

MURA Dataset Licensing Information

More information about the license can be found here. Deep Lake users may have access to a variety of publicly available datasets. We do not host or distribute these datasets, vouch for their quality or fairness, or claim that you have a license to use the datasets. It is your responsibility to determine whether you have permission to use the datasets under their license. If you’re a dataset owner and do not want your dataset to be included in this library, please get in touch through a GitHub issue. Thank you for your contribution to the ML community!

MURA Dataset Citation Information

				
					@article{rajpurkar2017mura,
  title={MURA: Large Dataset for Abnormality Detection in Musculoskeletal Radiographs},
  author={Rajpurkar, Pranav and Irvin, Jeremy and Bagul, Aarti and Ding, Daisy and Duan, Tony and Mehta, Hershel and Yang, Brandon and Zhu, Kaylie and Laird, Dillon and Ball, Robyn L and others},
  journal={arXiv preprint arXiv:1712.06957},
  year={2017}
}