Virtual Spectral Decomposition with Dendritic Tile Selection: An Explainable AI Framework for Multimodal Tissue Composition Analysis and Immune Phenotyping Across Pancreatic, Lung, and Breast Cancer
该论文提出了一种名为“虚拟光谱分解(VSD)”的可解释人工智能框架,通过模态无关的六通道组织成分分解和树突状瓦片选择算法,在胰腺、肺和乳腺癌的多模态影像分析中实现了无需分子检测即可进行可验证的肿瘤免疫表型分型与早期检测。