Principle-Guided Supervision for Interpretable Uncertainty in Medical Image Segmentation
For medical professionals relying on segmentation predictions, this work provides uncertainty estimates that are more interpretable and trustworthy, addressing a key limitation in high-stakes decision making.
The paper tackles the problem of interpretable uncertainty in medical image segmentation, ensuring that uncertainty estimates align with human-understandable sources of ambiguity. Their method, PriUS, produces uncertainty maps that are more consistent with image contrast, corruption severity, and geometric complexity compared to state-of-the-art methods, while maintaining competitive segmentation performance.
Uncertainty quantification complements model predictions by characterizing their reliability, which is essential for high-stakes decision making such as medical image segmentation. However, most existing methods reduce uncertainty to a scalar confidence estimate, leaving its spatial distribution semantically underconstrained. In this work, we focus on uncertainty interpretability, namely, whether estimated uncertainty behaves in a human-understandable manner with respect to sources of ambiguity. We identify three perception-aligned principles requiring the spatial distribution of uncertainty to reflect: (1) image contrast between structures, (2) severity of image corruption, and (3) geometric complexity in anatomical structures. Accordingly, we develop a principle-guided uncertainty supervision framework (PriUS) based on evidential learning, in which the corresponding supervision objectives are explicitly enforced during training. We further introduce quantitative metrics to measure the consistency between predicted uncertainty and image attributes that induce ambiguity. Experiments on ACDC, ISIC, and WHS datasets showed that, compared with state-of-the-art methods, PriUS produced more consistent uncertainty estimates while maintaining competitive segmentation performance.