From imaging to omics: deep learning is bridging MRI and liquid biopsy in bone tumor diagnosis

Bone tumors such as osteosarcoma and Ewing sarcoma remain among the most challenging cancers to diagnose and monitor because of their biological heterogeneity and overlapping radiological features. Magnetic resonance imaging (MRI) provides detailed anatomical insights, whereas liquid biopsy offers minimally invasive access to tumor genetics through circulating DNA, RNA, and extracellular vesicles. Each modality alone is limited, but recent advances in deep learning have enabled multimodal fusion of imaging and molecular data, improving risk stratification, therapy monitoring, and prognostication in patients with osteosarcoma and Ewing sarcoma. This review highlights how multimodal AI frameworks are being applied to bone tumors, delineating evidence from sarcoma-specific studies and representative pan‑cancer models with direct methodological relevance. By integrating MRI radiomics with liquid biopsy omics, deep learning holds promise for redefining precision oncology in bone tumors, delivering earlier detection and more personalized treatment strategies.