Knee Osteoarthritis Detection and Classification Using X-Rays

Knee osteoarthritis is a common form of arthritis, chronic and progressive disease recognized by joint space narrowing, osteophyte formation, sclerosis, bone deformity that can be observed using radiographs. Radiography regarded as the gold standard cheapest most readily available modality. X-ray images are graded Kellgren Lawrence’s (KL) grading scheme according to order severity from normal severe. Early detection help early treatment hence slows down knee degeneration. Unfortunately, existing approaches either merge or exclude perplexing grades improve performance their models. This study aims automatically detect classify KL system for We have proposed an automated deep learning-based ordinal classification approach diagnosis single posteroanterior standing x-ray image. An Osteoarthritis Initiative(OAI) based dataset chosen this study. The was split into training, testing, validation set with 7: 2: 1 ratio. took advantage transfer learning fine-tuned ResNet-34, VGG-19, DenseNet 121, 161 joined them in ensemble model’s overall performance. Our method has shown promising results obtaining 98% accuracy 0.99 Quadratic Weighted Kappa 95% confidence interval. Also, per grade significantly improved. Furthermore, our methods outperform state-of-the-art methods.