Knee Osteoarthritis Dataset with Severity
Dataset of radiographic images of the knee annotated according to five levels of osteoarthritis severity (grade 0 to 4), used for automatic classification and detection.
Approximately 9,786 annotated X-ray images in JPEG/DICOM format, with KL grades 0 to 4
Attribution 4.0 International (CC BY 4.0)
Description
The Knee Osteoarthritis Dataset with Severity contains nearly 10,000 X-ray images of the knee, annotated according to the KL system (Kellgren-Lawrence) with 5 grades ranging from healthy to severe. This dataset is designed for the automatic detection and evaluation of the severity of osteoarthritis using computer vision methods.
What is this dataset for?
- Develop models to aid medical diagnosis in orthopedic imaging
- Classifying the severity of knee osteoarthritis on X-ray images
- Contribute to research in medical artificial intelligence and computer vision
Can it be enriched or improved?
This dataset can be enriched by additional annotations (precise location of the lesions, segmentations). It is also possible to add associated clinical data to improve predictive models.
🔎 In summary
🧠 Recommended for
- Medical AI researchers
- Radiologists
- Health data scientists
🔧 Compatible tools
- PyTorch
- TensorFlow
- MONAI
- Medical annotation tools
- ITK-SNAP
💡 Tip
Use transfer learning techniques to optimize the classification with this dataset.
Frequently Asked Questions
What is the classification system used to annotate images?
The Kellgren-Lawrence (KL) system with 5 grades, from 0 (healthy) to 4 (severe osteoarthritis).
Can this dataset be used for commercial purposes?
Yes, provided that the CC BY 4.0 license is respected, in particular by correctly attributing the source.
Does the dataset contain personally identifiable data?
No, the images are anonymized and do not contain personal data.
