Generalisation effects of predictive uncertainty estimation in deep learning for digital pathology

TL;DR

This study evaluates how uncertainty estimation methods in deep learning improve reliability and misprediction detection in digital pathology, especially under domain shifts, with deep ensembles showing the best performance.

Contribution

It compares model-integrated and model-agnostic uncertainty estimation methods in digital pathology, highlighting their impact on robustness and misprediction detection.

Findings

Uncertainty estimates reduce sensitivity to classification thresholds.

Deep ensembles detect 70-90% of mispredictions.

Deep ensembles outperform other methods in robustness.

Abstract

Deep learning (DL) has shown great potential in digital pathology applications. The robustness of a diagnostic DL-based solution is essential for safe clinical deployment. In this work we evaluate if adding uncertainty estimates for DL predictions in digital pathology could result in increased value for the clinical applications, by boosting the general predictive performance or by detecting mispredictions. We compare the effectiveness of model-integrated methods (MC dropout and Deep ensembles) with a model-agnostic approach (Test time augmentation, TTA). Moreover, four uncertainty metrics are compared. Our experiments focus on two domain shift scenarios: a shift to a different medical center and to an underrepresented subtype of cancer. Our results show that uncertainty estimates increase reliability by reducing a model's sensitivity to classification threshold selection as well as by detecting between 70\% and 90\% of the mispredictions done by the model. Overall, the deep ensembles method achieved the best performance closely followed by TTA.