Fracture X-Ray Multi-Region
This dataset contains more than 10,000 radiographs classified as fractured or not, covering numerous anatomical regions (limbs, hips, spine...). It is designed for medical AI projects in fracture detection.
10,580 X-ray images (JPG), divided by train/test/validation
Open Data Commons Attribution License (ODC-by)
Description
The dataset Fracture X-Ray Multi-Region includes 10,580 X-ray images classified into two categories: fractured and unfractured. The data covers the main anatomical areas: upper and lower extremities, hips, knees, spine, etc. They are divided into training, validation, and test sets.
What is this dataset for?
- Forming binary classification models in radiology (fracture vs non-fracture)
- Testing CNN architectures on medical images
- Create assisted diagnostic tools for health professionals
Can it be enriched or improved?
Yes. It is possible to add more detailed annotations (location of fractures, type of fracture), to apply data augmentation techniques or to cross with other games to create more robust pipelines. This dataset can also be used for segmentation or multi-category classification with enrichment.
🔎 In summary
🧠 Recommended for
- Medical diagnostic projects
- Supervised classification
- AI imaging students
🔧 Compatible tools
- PyTorch
- TensorFlow
- FastAI
- MONAI
💡 Tip
For better performance, normalize contrasts and apply noise reduction in preprocessing.
Frequently Asked Questions
Can this dataset be used for commercial purposes?
Yes, subject to compliance with the terms of the ODC-by license, including attribution to the original authors.
Does this dataset contain accurate annotations of fractured areas?
No, labels are limited to a global classification (fractured or not), with no precise location.
Is it suitable for the automatic detection of fractures?
Yes, it can be used as a solid foundation for training automated medical detection or triage models.
