Back to all papers

A metaheuristic-based ensemble framework for osteoporosis classification using knee X-ray images.

November 17, 2026pubmed logopapers

Authors

Yelamati S,Thota S

Affiliations (3)

  • Research Scholar, Department of CSE, GITAM University, Rudraram, Andhra Pradesh, 502329, India. [email protected].
  • Vignan's Institute of Information Technology, Andhra Pradesh, India. [email protected].
  • Department of CSE, GITAM University, Rudraram, Andhra Pradesh, 502329, India.

Abstract

Osteoporosis is a progressive skeletal disorder characterized by reduced bone strength and increased fracture susceptibility, while its early identification from conventional radiographic images remains challenging because clinically relevant structural and textural variations can be subtle. This study proposes OsteoEnsembleNet, an ensemble stacking machine learning framework for automated osteoporosis classification using knee radiographs. The framework employs an Adaptive Unscented Kalman Filter (AUKF) for adaptive image quality enhancement, followed by a Hybrid Improved Beluga Whale Optimizer-Modified Fuzzy C-Means (HiBWmFCM) method for knee region segmentation. Multiple complementary feature domains, including morphological, statistical, gray-level co-occurrence matrix (GLCM), texture, and local binary pattern (LBP) descriptors, are subsequently extracted and optimized using HRSWaO feature selection. To improve classification robustness, predictions from seven heterogeneous classifiers-artificial neural network (ANN), k-nearest neighbors (KNN), support vector machine (SVM), random forest (RF), gradient boosting, XGBoost, and AdaBoost-are integrated through a logistic regression-based stacking meta-classifier. Experiments were conducted on publicly available Knee X-ray Osteoporosis Database. On the 80:20 patient-level split, the proposed framework achieved 97.92% accuracy, 98.96% precision, 96.67% recall, 97.72% F1-score, 96.01% MCC, and 95.63% Jaccard index, with a Hamming loss of 2.08%. The results demonstrate that combining adaptive preprocessing, targeted segmentation, complementary radiographic feature representation, feature selection, and heterogeneous ensemble learning can provide consistent discrimination among Normal, Osteopenia, and Osteoporosis categories from knee X-ray images. The framework is intended as an automated image-based decision-support approach and does not directly replace quantitative bone density assessment.

Topics

Journal Article

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAISlice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.