OsteoNet-X: A deep residual learning framework for automated osteoporosis detection from knee X-ray images
- 1School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, India
Abstract
Osteoporosis is a progressive skeletal disorder characterized by reduced bone mineral density and increased susceptibility to fractures, making early diagnosis essential for effective clinical intervention and long-term patient management. Recent advances in artificial intelligence have demonstrated significant potential for automated bone disease assessment using medical imaging modalities. In this study, a unified deep learning framework, termed OsteoNet-X, is proposed for automated osteoporosis detection and multi-class bone health classification using knee X-ray images. Unlike conventional image classification approaches, OsteoNet-X integrates YOLO-based Region of Interest (ROI) localization, DenseNet-based feature extraction, Feature-Aware Learning (FAL) optimization, and Grad-CAM-based Explainable Artificial Intelligence (XAI) to enhance both classification performance and model interpretability. The ROI extraction stage automatically identifies anatomically relevant knee joint regions, reducing background interference and enabling more discriminative feature learning. Subsequently, DenseNet facilitates efficient feature propagation and hierarchical representation learning for distinguishing normal, osteopenia, and osteoporosis conditions. The incorporation of XAI further provides visual explanations of model decisions by highlighting clinically relevant regions within the knee joint. Experimental evaluation demonstrates that OsteoNet-X achieves a classification accuracy of 97.08%, a macro F1-score of 96.61%, and a weighted F1-score of 97.11%.
Related
- Evidence for a cohesion-related function of alpaca hums: Linking automated call detection with GPS data
- An embedded deep learning–fuzzy logic framework for intelligent and automated irrigation in smart farming
- Efficient primer design for broad-range virus detection: A semi-automated workflow using sequence clustering and varVAMP
- An automated pipeline for adult and developmental neuropsychological assessment of executive functions and metacognition based on a multi-layer neuropsychological digital-twin framework
- Reproducible detection of antigen-specific T cells and Tregs via standardized and automated activation-induced marker assay workflows
BibTeX
@article{sangula2026osteonetx,
title = {{OsteoNet-X: A deep residual learning framework for automated osteoporosis detection from knee X-ray images}},
author = {Sudhir Kumar Sangula and S. P. Siddique Ibrahim},
journal = {PLOS One},
year = {2026},
month = oct,
doi = {10.1371/journal.pone.0361033},
url = {https://doi.org/10.1371/journal.pone.0361033},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - OsteoNet-X: A deep residual learning framework for automated osteoporosis detection from knee X-ray images
AU - Sudhir Kumar Sangula
AU - S. P. Siddique Ibrahim
PY - 2026
DA - 2026/10/27
JO - PLOS One
DO - 10.1371/journal.pone.0361033
UR - https://doi.org/10.1371/journal.pone.0361033
AB - Osteoporosis is a progressive skeletal disorder characterized by reduced bone mineral density and increased susceptibility to fractures, making early diagnosis essential for effective clinical intervention and long-term patient management. Recent advances in artificial intelligence have demonstrated significant potential for automated bone disease assessment using medical imaging modalities. In this study, a unified deep learning framework, termed OsteoNet-X, is proposed for automated osteoporosis detection and multi-class bone health classification using knee X-ray images. Unlike conventional image classification approaches, OsteoNet-X integrates YOLO-based Region of Interest (ROI) localization, DenseNet-based feature extraction, Feature-Aware Learning (FAL) optimization, and Grad-CAM-based Explainable Artificial Intelligence (XAI) to enhance both classification performance and model interpretability. The ROI extraction stage automatically identifies anatomically relevant knee joint regions, reducing background interference and enabling more discriminative feature learning. Subsequently, DenseNet facilitates efficient feature propagation and hierarchical representation learning for distinguishing normal, osteopenia, and osteoporosis conditions. The incorporation of XAI further provides visual explanations of model decisions by highlighting clinically relevant regions within the knee joint. Experimental evaluation demonstrates that OsteoNet-X achieves a classification accuracy of 97.08%, a macro F1-score of 96.61%, and a weighted F1-score of 97.11%.
SN - 1932-6203
ER -