Open Access

ARTICLE

Automated Severity Classification of Knee Osteoarthritis from Radiographs Using Transfer Learning Based Deep Neural Networks

Syed Nisar Hussain Bukhari^*, Sehar Altaf

National Institute of Electronics and Information Technology (NIELIT), Srinagar, India

* Corresponding Author: Syed Nisar Hussain Bukhari. Email:

Journal on Artificial Intelligence 2026, 8, 137-152. https://doi.org/10.32604/jai.2026.077943

Received 19 December 2025; Accepted 28 January 2026; Issue published 11 March 2026

Abstract

Knee osteoarthritis is a progressive degenerative joint disorder that leads to pain, stiffness, and reduced mobility, significantly affecting quality of life. Early and reliable diagnosis is essential for effective disease management, yet conventional radiographic assessment remains time-consuming and subject to inter-observer variability. This study presents a comparative deep learning (DL) based approach for automated severity classification of knee osteoarthritis using plain radiographic images. Multiple pretrained convolutional neural network architectures, including EfficientNetB3, InceptionNet, VGG19, ResNet, and EfficientNetV2S, were evaluated within a transfer learning paradigm. All models were trained and assessed on a publicly available dataset to classify knee osteoarthritis severity into clinically relevant categories. Among the evaluated architectures, EfficientNetB3 demonstrated the most consistent performance, achieving an accuracy of 0.97. Statistical significance analysis further confirmed that the performance differences between EfficientNetB3 and the other models were significant. The results indicate that modern DL architectures can provide reliable and consistent severity assessment, supporting their potential use as clinical decision support tools for knee osteoarthritis diagnosis.

---

Keywords

---