Knee Osteoarthritis Detection and Classification Using a Customized CenterNet with DenseNet201
H.Srikanth D.Srikar
Department of CSE (AI&ML) Department of CSE (AI&ML) 2111cs020556@mallareddyuniversity.ac.in 2111cs020557@mallareddyuniversity.ac.in
P.Srikar R.Srilekha
Department of CSE (AI&ML) Department of CSE(AI&ML) 2111cs020558@mallareddyuniversity.ac.in 2111cs020559@mallareddyuniversity.ac.in
D.Srinath
Department of CSE (AI&ML) 2111cs020560@mallareddyuniversity.ac.in
Prof P.Bhavani Assistant Professor Department
of AI & ML
MALLA REDDY UNIVERSITY
HYDERABAD
I. INTRODUCTION
Abstract: - The Knee osteoarthritis (OA) is a prevalent musculoskeletal disorder that significantly impacts quality of life. Early detection and classification of OA stages are crucial for effective management and treatment. This study proposes a novel approach for the automated detection and classification of knee osteoarthritis using a customized version of the CenterNet object detection model combined with DenseNet201 for feature extraction. The CenterNet model is adapted to detect and localize knee joints, while DenseNet201, a powerful convolutional neural network (CNN), is leveraged for its deep feature extraction and high efficiency in processing medical images. The proposed model is trained and validated on a dataset of knee X-ray images, categorizing them into various OA stages, including normal, mild, moderate, and severe. The integration of DenseNet201 into CenterNet improves the model’s ability to capture finegrained details in knee joint structures, enhancing classification accuracy. Experimental results show that the proposed method achieves a high classification performance compared to traditional image-based methods, demonstrating the potential of deep learning techniques in assisting clinicians with early OA detection and monitoring.
Keywords: Machine learning, detection performance, HCI, classification, deep learning, multi-scale features.
