Uncertainty Estimation in Pathology Foundation Models via Deep Mutual Learning

2026-06-29

AI Summary

This work addresses the limited reliability of existing pathology foundation models in clinical settings, where trustworthy confidence estimation is lacking and single models struggle to generalize across diverse tasks. To overcome these challenges, we propose DICE, a novel framework that, for the first time, applies deep mutual learning to ensembles of frozen pathology foundation models. DICE leverages inter-model disagreement as a proxy for uncertainty and enhances estimation quality through consistency alignment. Theoretical analysis demonstrates that our approach provides an upper bound on model uncertainty and enables unsupervised anomaly localization. Extensive experiments on three whole-slide image benchmarks show that DICE matches or surpasses state-of-the-art methods in in-distribution and out-of-distribution high-risk prediction identification, classification, calibration, and local anomaly detection.

Abstract

Pathology foundation models (PFMs) offer generalizable representations for whole-slide image (WSI) analysis, yet their clinical adoption remains limited. Specifically, their predictions lack reliable confidence estimates, and no single PFM is universally best across tasks, which severely undermines trust in medical settings. To overcome this, we propose $\mathtt{DICE}$, a plug-and-play framework that ensembles $K$ frozen PFMs and models their disagreement as a proxy for uncertainty estimation. To ensure this proxy yields meaningful estimates, we align the ensemble members via deep mutual learning, and theoretically show that this objective upper-bounds the model uncertainty. Additionally, we demonstrate that the ensemble's consensus localizes abnormalities at the patch level without any explicit supervision. We evaluate $\mathtt{DICE}$ on three challenging WSI benchmarks. Notably, our framework provides reliable uncertainty estimates that accurately flag failure-prone cases under in- and out-of-distribution settings, while matching or outperforming SOTA baselines in classification, calibration, and localization. Overall, $\mathtt{DICE}$ takes a crucial step toward translating PFMs into uncertainty-aware decision-support systems.

Problem

• Uncertainty Estimation
• Pathology Foundation Models
• Whole-slide Image Analysis
• Model Confidence
• Clinical Trust

Innovation

• Uncertainty Estimation
• Pathology Foundation Models
• Deep Mutual Learning
• Ensemble Learning
• Whole-Slide Image Analysis