Bone Fracture Diagnosis with Two-Stream CNNs: Integrating Multimodal Data and Explainable AI for Enhanced Clinical Decision-Making
Keywords:
Bone Fracture Diagnosis, Two-Stream CNN, Multimodal Data Fusion, Explainable AI, Clinical Metadata Integration, Radiographic Imaging, Grad-CAM Visualization, Machine Learning in Healthcare, Diagnostic Accuracy, Automated Medical ImagingAbstract
Accurate and timely bone fracture diagnosis is a critical challenge in medical imaging, demandingadvanced solutions to reduce false positives, improve interpretability, and handle multimodal data. This studyproposes a novel Two-Stream Compare and Contrast Network (TSCNN) integrating radiographic imaging and clinical metadata to address these challenges. Using the MURA dataset and a custom
References
[1] Sivapriya T, K. R. Sri Preethaa, Yuvaraj Natarajan, M. Shyamala Devi, "Multi‐Channel Fusion Residual Network for Robust Bone Fracture Classification From Radiographs", IET Image Processing, vol.20, no.1, 2026.
[2] kash Roy, Md. Istiak Hasan, Mst. Shikha Moni, Aminun Nahar, Shahnaj Parvin, "Enhancing Bone Fracture Identification with YOLOv11 andRobust KFold Cross-Validation", 2025 IEEE 4th International Conference on Robotics, Automation, Artificial-Intelligence and Internet-of-Things(RAAICON), pp.94-97, 2025
