Explainable AI in osteosarcoma image analysis: A deep learning and visualization approach

Osteosarcoma represents one of the most aggressive bone malignancies, predominantly affecting adolescents and young adults. Traditional diagnostic approaches rely heavily on radiological imaging and histopathological examination, which are time-intensive and subject to inter-observer variability. This research explores the integration of explainable artificial intelligence (XAI) techniques with deep learning models to enhance osteosarcoma detection and diagnosis through medical image analysis. We developed a convolutional neural network architecture combined with gradient-weighted class activation mapping (Grad-CAM) and Local Interpretable Model-agnostic Explanations (LIME) to provide transparent, clinically interpretable predictions. Our model was trained on 1847 radiographic images from multiple medical centers, achieving a classification accuracy of 94.3% with an AUC of 0.967. The XAI components successfully highlighted tumor regions with 89.7% concordance with expert radiologist annotations. This study demonstrates that explainable AI frameworks can bridge the gap between computational accuracy and clinical trust, offering radiologists a powerful assistive tool while maintaining diagnostic transparency. The findings suggest significant potential for reducing diagnostic time by approximately 40% while improving consistency in osteosarcoma identification across different clinical settings.