Charge Au@Pt NPs combined with 3D STS-Net for adaptive and sensitized radiotherapy of hepatocellular carcinoma: Synergistic enhancement of therapeutic gain across physical and biological dimensions

The therapeutic gain ratio (TGR) of radiotherapy for hepatocellular carcinoma (HCC) remains limited by two major barriers: insufficient precision in adaptive radiotherapy (ART) on the physical dimension and the lack of effective radiosensitization on the biological dimension. Although advances have been made separately in accurate dose delivery and tumor-sensitizing strategies, no approach has yet integrated both dimensions to achieve a coordinated improvement in TGR, representing a critical gap in current practice. In this study, we propose an integrated physical–biological strategy that combines nanomaterials with artificial intelligence (AI). We first constructed charge-engineered gold–platinum nanoparticles that respond to the acidic tumor microenvironment and enable prolonged, high-contrast computed tomography imaging of HCC. These enhanced images were then used to develop the first Transformer–convolutional neural network hybrid model (3D STS-Net) tailored for this scenario, enabling high-accuracy three-dimensional segmentation of small HCC for image-guided adaptive radiotherapy. In parallel, we systematically evaluated the nanoparticles’ radiosensitizing effects in vitro and in vivo. The nanoparticles provided stable imaging enhancement for up to 120 h and markedly improved tumor–liver contrast. The 3D STS-Net achieved high segmentation accuracy, supporting more precise contouring for HCC ART. Moreover, the nanoparticles significantly increased radiation-induced reactive oxygen species and enhanced tumor control in animal models. Together, these findings demonstrate that the proposed strategy simultaneously strengthens radiotherapy performance in both physical and biological dimensions, leading to a coordinated improvement in TGR. This integrated “nanomaterial–AI” framework offers a systematic and generalizable approach for enhancing radiotherapy effectiveness in HCC. Graphical Abstact: Charge Au@Pt NPs were constructed for enhanced in vivo CT imaging, integrated with AI-based automatic tumor segmentation, and applied to evaluate radiotherapy sensitization. • Charge Au@Pt NPs with acid-responsive aggregation achieve sustained high-contrast CT HCC imaging for up to 120 h. • The 3D STS-Net AI model, built on nanoparticle-enhanced imaging, overcomes 3D automatic segmentation bottlenecks for HCC lesions. • The system integrates imaging enhancement, intelligent segmentation, and radiotherapy sensitization for coordinated physical and biological optimization. • Synergy of nanomaterials and AI significantly improves radiotherapy therapeutic gain ratio (TGR). • A first-in-class “physical adaptation + biological sensitization” strategy provides a theoretical and practical framework for precision radiotherapy of HCC and other solid tumors.