Dynamic Hypergraph Representation for Bone Metastasis Cancer Analysis

TL;DR

This paper addresses the challenge of analyzing bone metastasis in WSIs by proposing DyHG, a dynamic hypergraph neural network that learns high-order relationships among tissue patches through a low-rank incidence and Gumbel-Softmax-based sampling. The method combines a dynamic hypergraph construction module with a hypergraph convolutional network and MIL-style attention pooling to produce WSI-level predictions for primary bone cancer origins and subtyping. Through extensive experiments on two large bone metastasis datasets and two public datasets, DyHG outperforms state-of-the-art MIL and graph-based baselines, while providing interpretable attention heatmaps and demonstrating favorable time efficiency and generalization. These results highlight the practical potential of dynamic hypergraphs for complex pathological analysis and precision oncology.

Abstract

Bone metastasis analysis is a significant challenge in pathology and plays a critical role in determining patient quality of life and treatment strategies. The microenvironment and specific tissue structures are essential for pathologists to predict the primary bone cancer origins and primary bone cancer subtyping. By digitizing bone tissue sections into whole slide images (WSIs) and leveraging deep learning to model slide embeddings, this analysis can be enhanced. However, tumor metastasis involves complex multivariate interactions with diverse bone tissue structures, which traditional WSI analysis methods such as multiple instance learning (MIL) fail to capture. Moreover, graph neural networks (GNNs), limited to modeling pairwise relationships, are hard to represent high-order biological associations. To address these challenges, we propose a dynamic hypergraph neural network (DyHG) that overcomes the edge construction limitations of traditional graph representations by connecting multiple nodes via hyperedges. A low-rank strategy is used to reduce the complexity of parameters in learning hypergraph structures, while a Gumbel-Softmax-based sampling strategy optimizes the patch distribution across hyperedges. An MIL aggregator is then used to derive a graph-level embedding for comprehensive WSI analysis. To evaluate the effectiveness of DyHG, we construct two large-scale datasets for primary bone cancer origins and subtyping classification based on real-world bone metastasis scenarios. Extensive experiments demonstrate that DyHG significantly outperforms state-of-the-art (SOTA) baselines, showcasing its ability to model complex biological interactions and improve the accuracy of bone metastasis analysis.

Figures (11)

• Figure 1: Hypergraph can capture the interactions between distant patches in a more direct way than conventional graph, where $p$ and $e$ denote patch and edge/hyperedge, respectively.
• Figure 2: The framework of DyHG. We first preprocess each WSI to obtain the initial embeddings of each patch, then dynamically construct a hypergraph through low-rank and Gumbel-Softmax-based sampling strategy. Next, hypergraph convolutional network is performed to obtain updated patch embeddings. Finally, we obtain the final prediction through global attention pooling.
• Figure 3: Data statistics of bone metastasis cancer data.
• Figure 4: Distribution of patch counts per WSI.
• Figure 5: Hyperparameters analysis on primary bone cancer origins classification task