IPGPhormer: Interpretable Pathology Graph-Transformer for Survival Analysis

2025-08-17

AI Summary

Current cancer survival analysis methods struggle to jointly model long-range spatial dependencies and capture local contextual information, while lacking intrinsic interpretability—limiting their clinical utility. To address this, we propose IPGPhormer, the first framework enabling dual-level (tissue- and cell-level) interpretability without requiring post-hoc annotations. IPGPhormer integrates graph neural networks with Transformer architectures within a multiple-instance learning paradigm to construct graph-structured representations of histopathological images, thereby jointly modeling global spatial relationships and local microenvironmental dependencies. Evaluated on four public datasets, IPGPhormer significantly outperforms state-of-the-art methods in both survival prediction accuracy and interpretability. The source code is publicly available, demonstrating strong potential for clinical decision support.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Computer Vision: Interpretability, Explainability, and Transparency
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Pathological images play an essential role in cancer prognosis, while survival analysis, which integrates computational techniques, can predict critical clinical events such as patient mortality or disease recurrence from whole-slide images (WSIs). Recent advancements in multiple instance learning have significantly improved the efficiency of survival analysis. However, existing methods often struggle to balance the modeling of long-range spatial relationships with local contextual dependencies and typically lack inherent interpretability, limiting their clinical utility. To address these challenges, we propose the Interpretable Pathology Graph-Transformer (IPGPhormer), a novel framework that captures the characteristics of the tumor microenvironment and models their spatial dependencies across the tissue. IPGPhormer uniquely provides interpretability at both tissue and cellular levels without requiring post-hoc manual annotations, enabling detailed analyses of individual WSIs and cross-cohort assessments. Comprehensive evaluations on four public benchmark datasets demonstrate that IPGPhormer outperforms state-of-the-art methods in both predictive accuracy and interpretability. In summary, our method, IPGPhormer, offers a promising tool for cancer prognosis assessment, paving the way for more reliable and interpretable decision-support systems in pathology. The code is publicly available at https://anonymous.4open.science/r/IPGPhormer-6EEB.

Problem

• Balancing long-range spatial and local contextual dependencies in survival analysis
• Lack of inherent interpretability in current pathological image analysis methods
• Improving predictive accuracy and interpretability in cancer prognosis assessment

Innovation

• Graph-Transformer models tumor microenvironment spatial dependencies
• Provides interpretability at tissue and cellular levels
• Outperforms state-of-the-art in accuracy and interpretability