Optimized Deep Transfer Learning For Knee Osteoarthritis Classification Using X-Ray Images

• K. Sudha,
• A. Rajiv kannan,

Vol. 19, No. 6, June 30, 2025
10.3837/tiis.2025.06.001, Download Paper (Free):

• Deep Learning
• Knee Osteoarthritis (KOA)
• LeNet
• Transfer learning (TL)
• Aquila Optimizer (AO)
• Fractional Tangent Search Aquila Optimizer (FTSAO)
• Fractional calculus (FC)
• U-net

Abstract

Knee Osteoarthritis (KOA) is a type of knee joint problem caused by decreased articular cartilage between the tibia and femur bones. It significantly affects people, particularly adults so that accurate and early diagnosis of KOA is crucial. This research implements an Enhanced Aquila Inspired (EAI) Deep transfer learning framework to classify KOA. An adaptive median filter is applied for pre-processing on the specific knee X-ray image. The area of interest is used to do joint space extraction. The U-Net is used for joint space segmentation, which is trained by Tangent Search Aquila Optimizer (TSAO). Transfer Learning (TL) is then used to classify KOA, and Convolutional Neural Network (CNN) is used with LeNet. The training of this model is done by Fractional Tangent Search Aquila Optimizer (FTSAO), which is the integration of Tangent Search Algorithm (TSA), Aquila Optimizer (AO), and Fractional calculus (FC). The resulting output is then classified as normal, impacted by grade 1 (G1), grade 2 (G2), grade 3 (G3), and grade 4 (G4). The proposed framework achieved high accuracy, true negative rate (TNR), and true positive rate (TPR) of 91.5%, 91.7%, and 91.3%, respectively. The results show that the proposed framework is used for the accurate classification of KOA, which offers a reliable tool for early diagnosis and progression assessment.

Statistics

Cite this article