Background and Aims: Nasopharyngeal carcinoma (NPC) are a rare epithelial malignancy with distinct geographic and etiological features, often associated with Epstein-Barr virus (EBV) infection. This narrative review synthesizes recent advancements in the application of machine learning (ML) to NPC, focusing on diagnostic imaging, tumor segmentation, classification, prognostication, and treatment planning.
Methods: The review analyzes literature pertaining to ML integration in oncology, emphasizing deep learning, convolutional neural networks, and radiomics in NPC-related imaging (MRI, CT, PET), histopathology, and biomarker analysis. This narrative review synthesizes and critically analyzes peer-reviewed literature from 2020 to 2025 pertaining to ML integration in oncology, emphasizing deep learning, convolutional neural networks, and radiomics in NPC-related imaging (MRI, CT, PET), histopathology, and biomarker analysis. It evaluates model performance, clinical applicability, and translational barriers.
Results: ML models demonstrate high accuracy in early detection (AUC > 0.90), tumor segmentation (Dice score 0.85–0.95), and staging (TNM concordance 85–95%). Integrating multimodal data-radiological, clinical, and molecular -enhances risk stratification and personalized radiotherapy. Challenges include dataset bias, limited model explainability, and infrastructural constraints in clinical deployment.
Conclusion: Machine learning holds transformative potential for improving the diagnosis, classification, and treatment planning of NPC. Future research should prioritize model generalizability, explainable AI, and equitable access, particularly in endemic regions. Advances in federated learning and multimodal integration are key to clinical translation.