Hierarchical Multi-Scale Graph Learning with Knowledge-Guided Attention for Whole-Slide Image Survival Analysis

TL;DR

This paper introduces HMKGN, a hierarchical graph network that models multi-scale spatial relationships in whole-slide images, significantly improving survival prediction accuracy in cancer cohorts.

Contribution

The paper presents a novel hierarchical multi-scale graph learning framework with spatial locality constraints for WSI-based survival analysis, outperforming existing MIL models.

Findings

Achieved 10.85% improvement in concordance index.

Statistically significant stratification of patient survival risk.

Validated on four TCGA cohorts with consistent performance.

Abstract

We propose a Hierarchical Multi-scale Knowledge-aware Graph Network (HMKGN) that models multi-scale interactions and spatially hierarchical relationships within whole-slide images (WSIs) for cancer prognostication. Unlike conventional attention-based MIL, which ignores spatial organization, or graph-based MIL, which relies on static handcrafted graphs, HMKGN enforces a hierarchical structure with spatial locality constraints, wherein local cellular-level dynamic graphs aggregate spatially proximate patches within each region of interest (ROI) and a global slide-level dynamic graph integrates ROI-level features into WSI-level representations. Moreover, multi-scale integration at the ROI level combines coarse contextual features from broader views with fine-grained structural representations from local patch-graph aggregation. We evaluate HMKGN on four TCGA cohorts (KIRC, LGG, PAAD, and STAD; N=513, 487, 138, and 370) for survival prediction. It consistently outperforms existing MIL-based models, yielding improved concordance indices (10.85% better) and statistically significant stratification of patient survival risk (log-rank p < 0.05).