Robust sensitivity control in digital pathology via tile score distribution matching

TL;DR

This paper presents a novel method for controlling sensitivity in digital pathology models, ensuring reliable deployment across different clinical settings with minimal calibration data.

Contribution

It introduces a sensitivity control technique using optimal transport and MIL, addressing distribution shifts and regulatory requirements in digital pathology.

Findings

Effective sensitivity control with few calibration samples

Validated across multiple cohorts and tasks

Improves robustness of pathology models

Abstract

Deploying digital pathology models across medical centers is challenging due to distribution shifts. Recent advances in domain generalization improve model transferability in terms of aggregated performance measured by the Area Under Curve (AUC). However, clinical regulations often require to control the transferability of other metrics, such as prescribed sensitivity levels. We introduce a novel approach to control the sensitivity of whole slide image (WSI) classification models, based on optimal transport and Multiple Instance Learning (MIL). Validated across multiple cohorts and tasks, our method enables robust sensitivity control with only a handful of calibration samples, providing a practical solution for reliable deployment of computational pathology systems.