Manuscript received May 5, 2026; revised June 2, 2026; accepted August 4, 2026; published November 14, 2026.

Abstract—Osteoarthritis (OA) is a chronic disorder of synovial joints characterized by deterioration of joint cartilage and underlying bone. The increasing prevalence of the disease, driven by an ageing population and rising obesity rates, necessitates the development of accurate diagnostic tools for early intervention. In this study, the Red Fox Optimization for Convolutional Neural Network (RFOCNN) framework was proposed to diagnose knee OA. It integrates the Red Fox Optimization (RFO) metaheuristic with a Convolutional Neural Network (CNN). Specifically, RFO optimizes key CNN hyperparameters. The RFOCNN is evaluated on the Digital Knee X-ray Images dataset (DKXI), which comprises 1820 knee X-ray images. It is designed to perform binary classification between normal knees and OA knees. Additional validation experiments compare the proposed framework with other Deep Learning (DL) models, metaheuristic methods, and alternative inputs. The RFOCNN framework achieved a testing accuracy of 90% and an Area Under the Curve (AUC) of 0.96. In comparisons against DL models, baseline CNN, CNN tuned with random search, and CNN optimized with other metaheuristic algorithms, the RFOCNN demonstrated a consistent performance advantage over the compared methods. Such results highlight that the RFOCNN framework can successfully identify knees with OA.

Keywords—Osteoarthritis, deep learning, red fox optimization

Cite: Abdullah Dawood Sulaiman, Raid Rafi Omar Al-Nima, and Tingting Han, "Diagnosing Osteoarthritis by Proposing Red Fox Optimization for Convolutional Neural Network" Journal of Image and Graphics, Vol. 14, No. 6, pp. 1015-1028, 2026.

Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).