Anchor-driven Multi-modal Multi-scale Expert Selection for Survival Prediction

Key points

• The integrative analysis of histopathological Whole-Slide Images (WSIs) and transcriptomic profiles holds significant promise for cancer survival prediction.1
• Specifically, we present an Anchor-driven Multi-modal Fusion (AMF) module, which introduces learnable semantic anchors as cross-modal mediators to bridge the semantic gap by enforcing a structurally regularized alignment between transcriptomic features and multi-scale pathology representations.1
• Built upon this aligned semantic space, we further design a Hierarchical Mixture-of-Experts (H-MoE) selection module to decouple the hierarchical prognostic selection process.1
• Extensive experiments on multiple TCGA cancer cohorts demonstrate that our AM$^2$ES achieves state-of-the-art performance while offering fine-grained interpretability by visualizing how specific molecular pathways drive the expert routing decisions across tissue scales.1

Sources (1)

• [1]Anchor-driven Multi-modal Multi-scale Expert Selection for Survival Prediction
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 03:32 AM

  To address these limitations, we propose an Anchor-driven Multi-modal Multi-scale Expert Selection (AM$^2$ES) framework for survival prediction.

  The integrative analysis of histopathological Whole-Slide Images (WSIs) and transcriptomic profiles holds significant promise for cancer survival prediction.

Extractive summary: sentences quoted from the sources.