Contrastive uncertainty learning for robust multi-structure segmentation in lower-pelvic MR

Bilal Sidiqi; Yipei Wang; Wen Yan; Shaheer U. Saeed; Daniel C. Alexander; Yipeng Hu

cs.CVOct 14, 2026

Contrastive uncertainty learning for robust multi-structure segmentation in lower-pelvic MR

Authors: Bilal Sidiqi, Yipei Wang, Wen Yan, Shaheer U. Saeed, Daniel C. Alexander, Yipeng Hu

Organizations: UCL Hawkes Institute, University College London, London, UK · Department of Medical Physics and Biomedical Engineering, University College London, London, UK · Centre for Bioengineering, School of Engineering and Materials Science, Queen Mary University of London, London, United Kingdom · Digital Environment Research Institute, Queen Mary University of London, London, United Kingdom ·

Abstract

Boundary delineation of anatomical structures in prostate MRI is often compromised by regional ambiguity. Automated segmentation methods relying on global overlap loss functions tend to be insensitive to surface deviations, leading to significant errors in ambiguous regions. Most existing uncertainty-aware methods treat uncertainty as a post-hoc quality indicator or prediction filter. We propose Contrastive Uncertainty Learning (CUL), a supervised framework that uses predictive uncertainty to directly guide representation learning. Using Monte Carlo Dropout, CUL identifies voxel-wise uncertainty and structures the embedding geometry accordingly: low-uncertainty, correctly predicted features form stable voxels while high-uncertainty or misclassified features are actively repelled, preventing feature collapse at difficult boundaries. We evaluate CUL on a 589-scan 3D prostate MRI multi-structure segmentation task (8 anatomical structures) using a ViT with UperNet decoder architecture. Average Dice ( $83.38\%$ ) remains comparable to the baseline ( $83.35\%$ ), while surface distance decreases significantly ( $p<0.001$ ) across 6 of 8 anatomical structures. A hard-case analysis reveals a $22{\times}$ asymmetry between significantly improved and degraded cases ( $18.8\%$ vs $0.9\%$ ). Ablation studies show that removing uncertainty-guided sample selection eliminates improvements for the transition zone, while replacing the contrastive objective with a boundary loss additionally removes bladder improvements, demonstrating that both uncertainty-guided sampling and the contrastive objective are essential for the observed gains. These findings demonstrate that the proposed uncertainty-guided representation learning is a useful mechanism for reducing the risk of extreme errors in safety-critical medical segmentation

Figures & tables

Figure 1: Overview of the proposed Contrastive Uncertainty Learning (CUL) framework. Prostate MRI is processed by a segmentation network to produce voxel wise predictions and an associated uncertainty map. Voxel embeddings are partitioned into low and high uncertainty regions and used to construct contrastive pairs. Low-uncertainty same-class features are pulled together, while high-uncertainty or misclassified features are repelled, improving boundary robustness.

Metric	Baseline	CUL	Class-Based	Classic BL
Dice ( $\uparrow$ )	$0.8335\pm 0.0647$	$0.8338\pm 0.0651$	$0.8331\pm 0.0637$	$\mathbf{0.8338\pm 0.0634}$
HD95(mm) ( $\downarrow$ )	$5.9343\pm 8.1414$	$\mathbf{5.4130\pm 4.8829}$	$5.3736\pm 4.6558$	$5.6253\pm 5.2306$
SD(mm) ( $\downarrow$ )	$1.6076\pm 2.5101$	$\mathbf{1.4190\pm 1.8135}$	$1.4385\pm 1.8873$	$1.4641\pm 1.9188$

Table 1: Overall segmentation performance for Baseline, CUL, and the two ablations. SD(mm) is significantly improved which is consistent with CUL’s design: predictive uncertainty concentrates at local boundary regions, so improvement manifests in surface-sensitive metrics rather than in global overlap (Dice) or gross deviation extremes (HD95(mm)).

Metric	Improved / Degraded (%)	Sig+ / Sig- (%)	Ratio
Dice ( $\uparrow$ )	64 / 53 (54.7 / 45.3)	–	–
HD95(mm) ( $\downarrow$ )	66 / 51 (56.4 / 43.6)	12 / 8 (10.3 / 6.8)	1.5 $\times$
SD(mm) ( $\downarrow$ )	79 / 38 (67.5 / 32.5)	22 / 1 (18.8 / 0.9)	22 $\times$

Table 2: Case-wise performance distribution (n=117). Significant improvement/degradation thresholds: SD

\pm

0.1 mm, HD95

\pm

0.5 mm.

Subset	n	$\Delta$ HD95 (mm)
Easy ( $<$ 4.0032)	58	$\approx 0$
Hard ( $\geq$ 4.0032)	59	$-1.1131$

Table 3: HD95(mm) change stratified by baseline difficulty (median=4.0032).

Panel of Figure 2 — Figure 2: Qualitative comparison between an improved case (top) and a degraded case (bottom). The top row highlights CUL’s role as geometric “insurance”, where it effectively corrects extreme boundary outliers in cases characterised by high regional ambiguity (e.g. central gland and both neurovascular bundles). CUL organises representations to stabilise boundaries by anchoring confident regions and isolating ambiguity. This visualisation demonstrates the framework’s capacity to correct significant spatial deviations while maintaining the integrity of stable predictions.

Structure	Baseline	CUL	Class-Based	Classic BL
Bladder	$1.17\pm 1.82$	$0.85^{***}$	$0.86^{*}$	$0.87$
Bone	$1.20\pm 1.94$	$1.03^{***}$	$1.02^{***}$	$1.01^{***}$
Obturator Int.	$1.43\pm 3.47$	$1.37$	$1.30$	$1.37$
Central Gland	$2.50\pm 13.4$	$1.89^{***}$	$2.10^{***}$	$2.02^{**}$
Transition Zone	$1.59\pm 1.13$	$1.50^{**}$	$1.55$	$1.58$
Rectum	$1.22\pm 0.98$	$1.15^{***}$	$1.14^{***}$	$1.13^{***}$

Table 4: Per-structure SD (mm

\downarrow

), n=117, paired Wilcoxon vs. Baseline. ∗

p{<}0.05

, ∗∗

p{<}0.01

, ∗∗∗

p{<}0.001

Medical Image Segmentation

Medical image segmentation delineates organs, lesions, and other structures in scans to support measurement, diagnosis, treatment planning, and research.

See all 786 papers

Principle-Guided Supervision for Interpretable Uncertainty in Medical Image Segmentation
May 9, 2026An Sui, Yuzhu Li, Gunter Schumann +2Image Segmentation Uncertainty Quantification
School of Data Science, Fudan University, Shanghai, China · Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China · National Heart and Lung Institute, Imperial College London, London, United Kingdom
SegWithU: Uncertainty as Perturbation Energy for Single-Forward-Pass Risk-Aware Medical Image Segmentation
Apr 16, 2026Tianhao Fu, Austin Wang, Charles Chen +2Image Segmentation Uncertainty Quantification
University of Toronto, Toronto, ON, Canada · Vector Institute, Toronto, ON, Canada · Project Neura, Toronto, ON, Canada +4
Trustworthy Medical Segmentation: Uncertainty-Aware U-Net Evaluation Under Clinical Image Degradation
Jul 22, 2026Pranav Kaliaperumal, Manisha KaliaperumalImage Corruption Robustness U-Net
Department of Computer Science University of Colorado Denver Aurora, CO, USA · Department of Biology (Pre-Medical) Creighton University Omaha, NE, USA

May 9, 2026J/K move · Enter open · S save

Principle-Guided Supervision for Interpretable Uncertainty in Medical Image Segmentation

An Sui, Yuzhu Li, Gunter Schumann, Fuping Wu+1

School of Data Science, Fudan University, Shanghai, China · Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, Shanghai, China · National Heart and Lung Institute, Imperial College London, London, United Kingdom

Image Segmentation Uncertainty Quantification Medical Image Segmentation Segmentation Uncertainty Estimation

Open paper AskPDF arXiv

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.

Abstract

Medical Image Segmentation

Explore similar work

Principle-Guided Supervision for Interpretable Uncertainty in Medical Image Segmentation

SegWithU: Uncertainty as Perturbation Energy for Single-Forward-Pass Risk-Aware Medical Image Segmentation

Trustworthy Medical Segmentation: Uncertainty-Aware U-Net Evaluation Under Clinical Image Degradation