Geometry-Aware State Space Model: A New Paradigm for Whole-Slide Image Representation

TL;DR

This paper introduces a geometry-aware hybrid hyperbolic-Euclidean model for whole-slide image analysis, improving tissue architecture and cellular morphology representation, and employs a structured state space model with a mixture-of-experts for enhanced classification.

Contribution

It proposes a novel dual geometric space embedding for WSIs, along with a structured sequence model and regional experts, advancing the accuracy of computational pathology methods.

Findings

Outperforms existing MIL methods on seven WSI datasets.

Effectively models hierarchical tissue structures and local morphology.

Enhances slide-level classification accuracy across multiple cancer types.

Abstract

Accurate analysis of histopathological images is critical for disease diagnosis and treatment planning. Whole-slide images (WSIs), which digitize tissue specimens at gigapixel resolution, are fundamental to this process but require aggregating thousands of patches for slide-level predictions. Multiple Instance Learning (MIL) tackles this challenge with a two-stage paradigm, decoupling tile-level embedding and slide-level prediction. However, most existing methods implicitly embed patch representations in homogeneous Euclidean spaces, overlooking the hierarchical organization and regional heterogeneity of pathological tissues. This limits current models' ability to capture global tissue architecture and fine-grained cellular morphology. To address this limitation, we introduce a hybrid hyperbolic-Euclidean representation that embeds WSI features in dual geometric spaces, enabling complementary modeling of hierarchical tissue structures and local morphological details. Building on this formulation, we develop BatMIL, a WSI classification framework that leverages both geometric spaces. To model long-range dependencies among thousands of patches, we employ a structured state space sequence model (S4) backbone that encodes patch sequences with linear computational complexity. Furthermore, to account for regional heterogeneity, we introduce a chunk-level mixture-of-experts (MoE) module that groups patches into regions and dynamically routes them to specialized subnetworks, improving representational capacity while reducing redundant computation. Extensive experiments on seven WSI datasets spanning six cancer types demonstrate that BatMIL consistently outperforms state-of-the-art MIL approaches in slide-level classification tasks. These results indicate that geometry-aware representation learning offers a promising direction for next-generation computational pathology.