Semantic-Anchored Evidential Fusion for Domain-Robust Whole-Slide Survival Analysis

2026-06-18

AI Summary

This study addresses the limited generalizability of existing whole-slide image (WSI) survival analysis methods across clinical centers, which stems from their reliance on pixel-level features highly sensitive to staining and scanning variations. To overcome this, the work introduces high-order pathological semantics as a domain-invariant representation, leveraging visual question answering (VQA) to extract semantic anchors. It proposes a dual-stream evidence extraction architecture combined with Dirichlet-based subjective logic to model uncertainty, enabling a cautious fusion of semantic and visual evidence. Evaluated in a zero-shot setting across four unseen clinical centers, the method achieves an average C-index improvement of 10.2% over current approaches. Moreover, semantic features exhibit significantly lower inter-center variability than pixel-level features, substantially enhancing cross-domain robustness and prediction reliability.

Abstract

Whole-slide images (WSIs) are widely used for computational cancer prognosis. However, most existing methods primarily focus on in-domain performance and fail to generalize across clinical centers. This limitation stems from their reliance on pixel-derived representations that are highly susceptible to domain-specific artifacts caused by staining protocols and scanner hardware. We hypothesize that high-level pathology semantics, such as tumor grade and micro-environmental architecture, provide a domain-invariant semantic representation that mirrors the robust diagnostic logic of human pathologists. Therefore, we propose a Semantic-Anchored Evidential Fusion Survival (SAEFS) framework, where SAEFS derives semantic anchors from WSIs via Visual Question Answering (VQA), employs a dual-stream WSI evidence extraction architecture, uses Dirichlet-based Subjective Logic to model uncertainty, and fuses semantic and visual evidence through a cautious conjunction rule to avoid overconfident fusion from correlated sources. Trained exclusively on one source domain and evaluated zero-shot across four unseen domains, SAEFS consistently outperforms state-of-the-art models both in prediction accuracy and reliability, improving the average C-index by 10.2%. Quantitative analyses further show that VQA-derived semantic features exhibit significantly lower cross-center divergence than pixel-derived features, highlighting their robustness for cross-center clinical applications.

Problem

• domain robustness
• whole-slide image
• survival analysis
• cross-domain generalization
• computational pathology

Innovation

• Semantic-Anchored Fusion
• Domain-Robust Survival Analysis
• Visual Question Answering (VQA)
• Evidential Reasoning
• Whole-Slide Image