Skip to content

Latest commit

 

History

History
117 lines (82 loc) · 5.75 KB

MURA.md

File metadata and controls

117 lines (82 loc) · 5.75 KB
Raw

MURA

Dataset Information

MURA is a musculoskeletal radiographic dataset that includes 14,863 studies of 12,173 patients, totaling 40,561 multi-view radiographic images. Each image is categorized into one of seven standard types of upper extremity radiographic studies: elbow, finger, forearm, hand, humerus, shoulder, and wrist. From 2001 to 2012, radiologists at Stanford Hospital manually annotated these studies during clinical radiographic interpretations, classifying them as normal or abnormal. The test set includes additional annotations by six radiologists from Stanford Hospital, covering 207 musculoskeletal studies. These radiologists reviewed and labeled each study in the test set as normal or abnormal individually using the PACS system in a clinical reading room environment. The radiologists had an average of 8.83 years of experience, ranging from 2 to 25 years. Three of these radiologists were randomly selected to create a gold standard, defined as the majority vote of these radiologists' labels.

The MURA anomaly detection task is a binary classification task where the input is an upper extremity radiographic study (containing one or more views/images), and the expected output is a binary label y ∈ {0,1}, indicating whether the study is normal or abnormal. Additionally, the MURA dataset serves as an educational tool in medical imaging, helping medical students, radiology technologists, and other learners to better understand and master the radiographic features of the musculoskeletal system, enhancing their clinical skills and diagnostic capabilities.

Dataset Meta Information

Dimensions Modality Task Type Anatomical Structures Anatomical Area Number of Categories Data Volume File Format
2D X-Ray Classification Musculoskeletal Muscle 2 40561 .png

Resolution Details

Dataset Statistics size
min (512, 108)
median (488, 400)
max (512, 268)

Label Information Statistics

Original dataset annotation information: Each image belongs to seven standard upper limb radiographic images, including elbow, finger, forearm, hand, humerus, shoulder, and wrist.

Study Train#Normal Train#Abnormal Validation#Normal Validation#Abnormal Total
Elbow 1094 660 92 66 1912
Finger 1280 655 92 83 2110
Hand 1497 521 101 66 2185
Humerus 321 271 68 67 727
Forearm 590 287 69 64 1010
Shoulder 1364 1457 99 95 3015
Wrist 2134 1326 140 97 3697
Total No. of Studies 8280 5177 661 538 14656

The statistical results of normal and abnormal images are subject to the actual statistics of the downloaded dataset:

Category train val
Normal bone X-ray 8280 661
Abnormal bone X-ray 5177 538

Visualization

Visualization results: normal elbow (upper left), abnormal fingers (upper right).

Visualization results.

File Structure

MURA
├── images
│   ├── train
│   │   ├── xxxx.png
│   │   ├── xxxx.png
│   │   ├── .....png
│   │   ├── xxxx.png
│   ├── valid
│   │   ├── xxxx.png
│   │   ├── xxxx.png
│   │   ├── .....png
│   │   ├── xxxx.png
├── train.txt
├── val.txt
├── README.md

Authors and Institutions

Pranav Rajpurkar (Stanford University)

Jeremy Irvin (Stanford University)

Aarti Bagul (Stanford Universit)

Daisy Ding (Stanford Universit)

Tony Duan (Stanford Universit)

Andrew Y. Ng (Stanford Universit)

Source Information

Official Website: https://stanfordmlgroup.github.io/competitions/mura/

Download Link: https://tianchi.aliyun.com/dataset/92011

Article Address: https://arxiv.org/pdf/1712.06957

Publication Date: 2018

Citation

@article{MURA
  title={MURA:LargeDatasetforAbnormalityDetectionin MusculoskeletalRadiographs},
  author = {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}
 }

Original introduction article is here.