标题： 组织学和转录组学的解耦多模态学习用于癌症表征 显示英文标题

标题： Disentangled Multi-modal Learning of Histology and Transcriptomics for Cancer Characterization

摘要： 组织病理学仍然是癌症诊断和预后的金标准。 随着转录组分析的出现，结合转录组学与组织学的多模态学习提供了更全面的信息。 然而，现有的多模态方法受到固有多模态异质性、多尺度整合不足以及对配对数据的依赖的挑战，限制了临床适用性。 为了解决这些挑战，我们提出了一种解耦的多模态框架，包含四个贡献：1）为减轻多模态异质性，我们使用解耦的多模态融合模块将WSI和转录组分解为肿瘤和微环境子空间，并引入一种基于置信度的梯度协调策略来平衡子空间优化。 2）为增强多尺度整合，我们提出了一种跨放大倍数的基因表达一致性策略，以对齐不同WSI放大倍数下的转录组信号。 3）为减少对配对数据的依赖，我们提出了一种子空间知识蒸馏策略，通过仅使用WSI的学生模型实现转录组无关的推理。 4）为提高推理效率，我们提出了一种信息标记聚合模块，在抑制WSI冗余的同时保留子空间语义。 在癌症诊断、预后和生存预测方面的大量实验表明，我们的方法在多种设置下均优于最先进的方法。 代码可在 https://github.com/helenypzhang/Disentangled-Multimodal-Learning 获取。 显示英文摘要

摘要： Histopathology remains the gold standard for cancer diagnosis and prognosis. With the advent of transcriptome profiling, multi-modal learning combining transcriptomics with histology offers more comprehensive information. However, existing multi-modal approaches are challenged by intrinsic multi-modal heterogeneity, insufficient multi-scale integration, and reliance on paired data, restricting clinical applicability. To address these challenges, we propose a disentangled multi-modal framework with four contributions: 1) To mitigate multi-modal heterogeneity, we decompose WSIs and transcriptomes into tumor and microenvironment subspaces using a disentangled multi-modal fusion module, and introduce a confidence-guided gradient coordination strategy to balance subspace optimization. 2) To enhance multi-scale integration, we propose an inter-magnification gene-expression consistency strategy that aligns transcriptomic signals across WSI magnifications. 3) To reduce dependency on paired data, we propose a subspace knowledge distillation strategy enabling transcriptome-agnostic inference through a WSI-only student model. 4) To improve inference efficiency, we propose an informative token aggregation module that suppresses WSI redundancy while preserving subspace semantics. Extensive experiments on cancer diagnosis, prognosis, and survival prediction demonstrate our superiority over state-of-the-art methods across multiple settings. Code is available at https://github.com/helenypzhang/Disentangled-Multimodal-Learning.