Search papers, labs, and topics across Lattice.
This paper introduces $\mathtt{DICE}$, a novel framework that ensembles multiple frozen pathology foundation models (PFMs) to improve uncertainty estimation in whole-slide image (WSI) analysis. By leveraging deep mutual learning to align the ensemble members, the authors demonstrate that the model's disagreement serves as a reliable proxy for uncertainty, enabling accurate identification of failure-prone cases. Evaluated on three challenging benchmarks, $\mathtt{DICE}$ not only provides robust uncertainty estimates but also matches or surpasses state-of-the-art performance in classification, calibration, and localization tasks.
Uncertainty estimates from pathology models can be reliably derived from ensemble disagreements, significantly enhancing trust in clinical applications.
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.
