VGG-16, VGG-16 With Random Forest, Resnet50 With SVM, and EfficientNetB0 with XGBoost-Enhancing Bone Fracture Classification in X-Ray Using Deep Learning Models

Bone fracture classification using X-ray images plays a vital role in clinical diagnosis and treatment planning. This study presents a comparative deep learning framework integrating four advanced models—VGG-16, VGG-16 with Random Forest, ResNet-50 with Support Vector Machine (SVM), and EfficientNetB0 with XGBoost—for accurate detection and classification of ten fracture types: avulsion, comminuted, fracture dislocation, greenstick, hairline, impacted, longitudinal, oblique, pathological, and spiral fractures. The VGG-16 network is fine-tuned for feature extraction and classification, while its Random Forest hybrid improves decision robustness by combining deep features with ensemble learning. ResNet-50 with SVM enhances feature discrimination through deep residual representations and optimized hyperplane separation. EfficientNetB0 with XGBoost further refines learning efficiency by integrating lightweight convolutional architectures with gradient-boosted decision trees. Experimental analysis on a labeled X-ray dataset demonstrates that hybrid models significantly outperform standalone CNNs, achieving higher precision, recall, and F1-scores. The integration of Random Forest, SVM, and XGBoost classifiers with deep features enhances interpretability and generalization. Overall, the proposed ensemble-based deep learning approach ensures robust, automated fracture classification, supporting faster and more reliable orthopedic diagnosis in clinical settings.

Keywords: Bone fracture, ensemble learning models, fracture classification, machine learning, medical imaging diagnostics, random forest, ResNet-50, VGG-16.