Pipeline of knee osteoarthritis grading pipeline, which includes knee joints detection and knee OA grading.
- Detecting two knee joints in X-ray images using a customized YOLOv2 model.
- Classifying the KL grade of detected knee joints with a novel ordinal loss.
Knee joint detection (DetJoint) and KL grading (ClsKL) training/testing datasets, as well as best models, can be downloaded from KneeXrayData, around 7G.
Please consider cite the paper if you use the code or data for your research.
@article{chen2019fully,
title={Fully Automatic Knee Osteoarthritis Severity Grading Using Deep Neural Networks with a Novel Ordinal Loss},
author={Chen, Pingjun and Gao, Linlin and Shi, Xiaoshuang and Allen, Kyle and Yang Lin},
journal={Computerized Medical Imaging and Graphics},,
volume={75},
pages={84--92},
year={2019},
doi={https://doi.org/10.1016/j.compmedimag.2019.06.002},
publisher={Elsevier}
}