Knee Osteoarthritis Detection and Classification Using a Customized CenterNet with DenseNet201
1 Dr. P. Anjaiah 2 V.Santosh,3 V.Ganesh,4 P.Jahnavi,5 G.Yashaswini
1 Assistant Professor, Department of Computer Science & Engineering (Artificial Intelligence & Machine Learning), Malla Reddy University, Kompally, Hyderabad. 1 Email : drmohammad.adambaba@mallareddyuniversity.ac.in
2,3,4,5 Students, Department of Computer Science & Engineering (Artificial Intelligence & Machine Learning), Malla Reddy University, Kompally, Hyderabad. 2 Email : 2211cs020460@mallareddyuniversity.ac.in, 3 Email: 2211cs020537@mallareddyuniversity.ac.in 4 Email: 2211cs020557@mallareddyuniversity.ac.in, 5 Email: 2211cs020581@mallareddyuniversity.ac.in
Abstract:
Knee Osteoarthritis (KOA) is a progressive degenerative joint disease that significantly impacts mobility and quality of life. Early detection and accurate severity grading are critical for effective clinical management and prevention of disease progression. Manual interpretation of knee X-ray images using the Kellgren–Lawrence (KL) grading system is time-consuming and subject to inter-observer variability. This research presents an end-to-end deep learning–based framework for automated detection and classification of knee osteoarthritis severity using transfer learning with MobileNetV2. The proposed system leverages a publicly available knee X-ray dataset categorized into five severity grades (0–4). To address class imbalance, class weighting strategies were incorporated during training. The final model achieved a validation accuracy of 54% with improved macro F1-score balance across all classes. Furthermore, the trained model was deployed using a Streamlit-based web interface to enable real-time image-based severity prediction. The system demonstrates the feasibility of lightweight deep learning architectures for practical medical image classification and clinical decision support.
Keywords: Knee Osteoarthritis, Deep Learning, Transfer Learning, MobileNetV2, Medical Image Classification, X-ray Analysis, Streamlit Deployment, Class Imbalance
