Evidential DualU-Net: Single-Pass Uncertainty for Cell Instance Segmentation

David Anglada-Rotger, Ferran Marques, Montse Pardàs
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:3103-3130, 2026.

Abstract

Accurate and trustworthy cell instance segmentation requires models that not only detect and classify nuclei but also communicate how much evidence supports each prediction. DualU-Net is a fast and effective two-head multi-task architecture for this problem, but—like most deterministic models—it provides no principled uncertainty estimates. We introduce Evidential DualU-Net, the first evidential framework for multi-task cell instance segmentation.Its segmentation head predicts Dirichlet concentration parameters, enabling single-pass, closed-form aleatoric, epistemic, and vacuity uncertainties at the pixel level, with instance-level quantities obtained via size-invariant pooling of pixel evidence. The centroid decoder is complemented with two lightweight geometric uncertainty cues that quantify localisation reliability without auxiliary models or sampling. Together, these evidential and geometric measures expose complementary failure modes and allow principled filtering of low-confidence nuclei. Across multi-tissue and multi-stain datasets, Evidential DualU-Net matches or surpasses deep ensembles and MC Dropout in error separation at a fraction of the cost, maintains or improves calibration over deterministic baselines, and generalises across datasets without retuning. This work provides an interpretable and computationally practical uncertainty formulation for digital pathology.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-anglada-rotger26a, title = {Evidential DualU-Net: Single-Pass Uncertainty for Cell Instance Segmentation}, author = {Anglada-Rotger, David and Marques, Ferran and Pard{\`a}s, Montse}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {3103--3130}, year = {2026}, editor = {Huo, Yuankai and Gao, Mingchen and Kuo, Chang-Fu and Jin, Yueming and Deng, Ruining}, volume = {315}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v315/main/assets/anglada-rotger26a/anglada-rotger26a.pdf}, url = {https://proceedings.mlr.press/v315/anglada-rotger26a.html}, abstract = {Accurate and trustworthy cell instance segmentation requires models that not only detect and classify nuclei but also communicate how much evidence supports each prediction. DualU-Net is a fast and effective two-head multi-task architecture for this problem, but—like most deterministic models—it provides no principled uncertainty estimates. We introduce Evidential DualU-Net, the first evidential framework for multi-task cell instance segmentation.Its segmentation head predicts Dirichlet concentration parameters, enabling single-pass, closed-form aleatoric, epistemic, and vacuity uncertainties at the pixel level, with instance-level quantities obtained via size-invariant pooling of pixel evidence. The centroid decoder is complemented with two lightweight geometric uncertainty cues that quantify localisation reliability without auxiliary models or sampling. Together, these evidential and geometric measures expose complementary failure modes and allow principled filtering of low-confidence nuclei. Across multi-tissue and multi-stain datasets, Evidential DualU-Net matches or surpasses deep ensembles and MC Dropout in error separation at a fraction of the cost, maintains or improves calibration over deterministic baselines, and generalises across datasets without retuning. This work provides an interpretable and computationally practical uncertainty formulation for digital pathology.} }
Endnote
%0 Conference Paper %T Evidential DualU-Net: Single-Pass Uncertainty for Cell Instance Segmentation %A David Anglada-Rotger %A Ferran Marques %A Montse Pardàs %B Proceedings of The 9th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Yuankai Huo %E Mingchen Gao %E Chang-Fu Kuo %E Yueming Jin %E Ruining Deng %F pmlr-v315-anglada-rotger26a %I PMLR %P 3103--3130 %U https://proceedings.mlr.press/v315/anglada-rotger26a.html %V 315 %X Accurate and trustworthy cell instance segmentation requires models that not only detect and classify nuclei but also communicate how much evidence supports each prediction. DualU-Net is a fast and effective two-head multi-task architecture for this problem, but—like most deterministic models—it provides no principled uncertainty estimates. We introduce Evidential DualU-Net, the first evidential framework for multi-task cell instance segmentation.Its segmentation head predicts Dirichlet concentration parameters, enabling single-pass, closed-form aleatoric, epistemic, and vacuity uncertainties at the pixel level, with instance-level quantities obtained via size-invariant pooling of pixel evidence. The centroid decoder is complemented with two lightweight geometric uncertainty cues that quantify localisation reliability without auxiliary models or sampling. Together, these evidential and geometric measures expose complementary failure modes and allow principled filtering of low-confidence nuclei. Across multi-tissue and multi-stain datasets, Evidential DualU-Net matches or surpasses deep ensembles and MC Dropout in error separation at a fraction of the cost, maintains or improves calibration over deterministic baselines, and generalises across datasets without retuning. This work provides an interpretable and computationally practical uncertainty formulation for digital pathology.
APA
Anglada-Rotger, D., Marques, F. & Pardàs, M.. (2026). Evidential DualU-Net: Single-Pass Uncertainty for Cell Instance Segmentation. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:3103-3130 Available from https://proceedings.mlr.press/v315/anglada-rotger26a.html.

Related Material