Modeling Whole-Slide Images as Dynamic Tumor Microenvironment Fields

2026-09-28

AI Summary

This study addresses the limitation that existing whole slide image (WSI) analysis neglects the spatial coherence and dynamic regional interactions within the tumor microenvironment (TME). To this end, we propose TMEvolve, a framework that models WSIs as dynamic microenvironment fields. It introduces a novel reaction-diffusion-inspired graph discretization evolution mechanism and incorporates concept-guided boundary flux to simulate signal propagation across heterogeneous regions. By integrating multiple instance learning, graph neural networks, and vision-language techniques, TMEvolve enables adaptive region formation and evolution. Extensive experiments on six datasets demonstrate that our method significantly outperforms mainstream baselines in tasks such as survival prediction, effectively enhancing both the accuracy and interpretability of weakly supervised computational pathology.

Abstract

Due to the gigapixel-scale nature of whole-slide images (WSIs), weakly supervised WSI analysis is commonly formulated as a multiple instance learning (MIL) problem, where patch-level features are aggregated into slide-level representations. However, diagnostic and prognostic evidence often arises from spatially coherent tumor microenvironment regions and their interactions, rather than isolated patches alone. Existing patch-level or static region-based methods usually overlook how tissue regions should be adaptively formed and subsequently evolved through microenvironment interactions across heterogeneous boundaries. In this paper, we propose Concept-Guided Tumor Microenvironment Evolution (TMEvolve), a reaction-diffusion-inspired framework that models WSIs as latent tumor microenvironment fields over discrete patch graphs. TMEvolve instantiates this view as a learnable graph-discretized evolution process over patch neighborhoods. It first forms adaptive soft tissue regions as coherent microenvironment units, then performs pseudo-time evolution through two complementary local dynamics: intra-region diffusion, which stabilizes latent states within coherent tissue compartments, and concept-guided boundary flux, which propagates visual feature signals and language-derived concept signals across heterogeneous region interfaces. The evolved microenvironment regions are finally aggregated for slide-level prediction. We evaluate TMEvolve on six datasets across three weakly supervised WSI tasks: survival prediction, gene expression prediction, and histological subtype classification. TMEvolve consistently improves over representative MIL methods, pathology foundation models, and concept-guided baselines. Ablation studies and visualizations further support the effectiveness and interpretability of TMEvolve, highlighting the value of dynamic region modeling and boundary interaction.

Problem

• Whole-Slide Images
• Tumor Microenvironment
• Multiple Instance Learning
• Dynamic Region Modeling
• Weakly Supervised Learning

Innovation

• Whole-Slide Images
• Tumor Microenvironment
• Reaction-Diffusion
• Multiple Instance Learning
• Concept-Guided Evolution