标题： MiCo：用于全切片图像分析的上下文感知聚类多重实例学习 显示英文标题

标题： MiCo: Multiple Instance Learning with Context-Aware Clustering for Whole Slide Image Analysis

摘要： 多实例学习（MIL）在组织病理学全切片图像（WSI）分析中用于癌症诊断和预后显示出显著的前景。 然而，WSI的固有空间异质性带来了关键挑战，因为形态学相似的组织类型通常分散在不同的解剖区域。 传统MIL方法难以建模这些分散的组织分布并有效捕捉跨区域的空间交互。 为了解决这些限制，我们提出了一种具有上下文感知聚类的新型多实例学习框架（MiCo），旨在增强WSI中的跨区域组织内相关性和加强组织间语义关联。 MiCo首先通过聚类实例来提炼有区别的形态模式，聚类中心作为语义锚点。 为了增强跨区域的组织内相关性，MiCo采用了一个聚类路径模块，通过特征相似性动态连接不同区域的同类型实例。 这些语义锚点作为上下文中心，传播语义关系以优化实例级表示。 为了消除语义碎片并加强组织间语义关联，MiCo集成了一个聚类缩减模块，该模块合并冗余锚点，同时增强不同语义组之间的信息交换。 在九个大规模公开癌症数据集上的两个具有挑战性的任务的广泛实验表明了MiCo的有效性，展示了其优于最先进方法的性能。 代码可在 https://github.com/junjianli106/MiCo 获取。 显示英文摘要

摘要： Multiple instance learning (MIL) has shown significant promise in histopathology whole slide image (WSI) analysis for cancer diagnosis and prognosis. However, the inherent spatial heterogeneity of WSIs presents critical challenges, as morphologically similar tissue types are often dispersed across distant anatomical regions. Conventional MIL methods struggle to model these scattered tissue distributions and capture cross-regional spatial interactions effectively. To address these limitations, we propose a novel Multiple instance learning framework with Context-Aware Clustering (MiCo), designed to enhance cross-regional intra-tissue correlations and strengthen inter-tissue semantic associations in WSIs. MiCo begins by clustering instances to distill discriminative morphological patterns, with cluster centroids serving as semantic anchors. To enhance cross-regional intra-tissue correlations, MiCo employs a Cluster Route module, which dynamically links instances of the same tissue type across distant regions via feature similarity. These semantic anchors act as contextual hubs, propagating semantic relationships to refine instance-level representations. To eliminate semantic fragmentation and strengthen inter-tissue semantic associations, MiCo integrates a Cluster Reducer module, which consolidates redundant anchors while enhancing information exchange between distinct semantic groups. Extensive experiments on two challenging tasks across nine large-scale public cancer datasets demonstrate the effectiveness of MiCo, showcasing its superiority over state-of-the-art methods. The code is available at https://github.com/junjianli106/MiCo.