Salvato in:
Dettagli Bibliografici
Autori principali: Sui, An, Li, Yuzhu, Schumann, Gunter, Wu, Fuping, Zhuang, Xiahai
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.10984
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911672080269312
author Sui, An
Li, Yuzhu
Schumann, Gunter
Wu, Fuping
Zhuang, Xiahai
author_facet Sui, An
Li, Yuzhu
Schumann, Gunter
Wu, Fuping
Zhuang, Xiahai
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2605_10984
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Principle-Guided Supervision for Interpretable Uncertainty in Medical Image Segmentation
Sui, An
Li, Yuzhu
Schumann, Gunter
Wu, Fuping
Zhuang, Xiahai
Computer Vision and Pattern Recognition
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.
title Principle-Guided Supervision for Interpretable Uncertainty in Medical Image Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.10984