Research Progress on Multi-Modal Imaging-Assisted Artificial Intelligence for Target Delineation and Treatment Outcome Prediction in Non–Small Cell Lung Cancer Radiotherapy

Non-small cell lung cancer (NSCLC) represents about 85% of total lung cancer cases. The main curative treatment for locally advanced patients who cannot undergo surgery relies on radiotherapy but target delineation using traditional CT faces challenges related to physician skill and lacks adequate predictive dimensions for treatment outcomes. The integration of artificial intelligence technologies with multimodal imaging has become more common to enhance both target delineation precision and treatment effectiveness prediction. The functional imaging capabilities of PET/CT and MRI help doctors evaluate suspicious metastatic lymph nodes and separate tumors from atelectasis or inflammation through metabolic and microstructural information. Deep learning algorithms enable radiomics models to extract image-based high-dimensional features which predict local control rates and survival outcomes and other clinical results. The current implementation of AI-assisted multimodal imaging in NSCLC radiotherapy faces multiple challenges including non-standardized data and inadequate model interpretability and insufficient clinical workflow integration. Further radiomics and multi-omics research needs urgent attention especially when radiotherapy combines with immunotherapy or targeted therapy. The future development requires three main steps: enhancing large-scale multicenter data sharing for external validation and developing interpretable AI algorithms and individualized adaptive radiotherapy strategies while promoting standardized clinical AI-assisted multimodal imaging applications to achieve precise and efficient NSCLC radiotherapy.