pith. sign in

arxiv: 2411.09553 · v4 · submitted 2024-11-14 · 💻 cs.CV

OOD-SEG: Exploiting out-of-distribution detection techniques for learning image segmentation from sparse multi-class positive-only annotations

Junwen Wang , Zhonghao Wang , Oscar MacCormac , Jonathan Shapey , Tom Vercauteren This is my paper

Pith reviewed 2026-05-23 17:42 UTC · model grok-4.3

classification 💻 cs.CV
keywords positive-unlabelled learningout-of-distribution detectionimage segmentationsparse annotationsmedical imagingsurgical imagingPU learning
0
0 comments X

The pith

Multi-class segmentation from sparse positive-only annotations works by applying OOD detection at the pixel level to treat unlabelled pixels as out-of-distribution.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out to reduce the annotation burden in medical image segmentation by training only on sparse positive labels for multiple classes and omitting background labels entirely. It casts the problem as pixel-wise positive-unlabelled learning and solves it by repurposing classification-style OOD detectors to flag unlabelled pixels that belong to background or unseen classes. The approach requires no background annotation at training time and produces models that can also identify OOD regions at test time. Evaluation uses a cross-validation scheme that holds out labelled classes as proxies for true OOD behaviour, with results reported on both hyperspectral and RGB surgical datasets.

Core claim

Multi-class segmentation with sparse positive-only annotations is formulated as a pixel-wise PU learning problem and addressed by integrating any OOD detection method at the pixel level, treating background together with unseen classes as the OOD set and forgoing background annotation.

What carries the argument

Pixel-level integration of OOD detection methods inside a positive-unlabelled learning framework for multi-class segmentation.

If this is right

  • Segmentation training no longer requires any background class labels.
  • Models can flag OOD pixels at deployment to reduce spurious outputs.
  • Any existing classification OOD detector can be dropped in at pixel resolution.
  • A cross-validation protocol that holds out classes serves as a proxy for real OOD evaluation in medical segmentation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same pixel-wise OOD framing could be tested on non-medical dense prediction tasks that also suffer from incomplete background labels.
  • Performance gains may depend on how well the chosen OOD scorer separates positive pixels from true background at fine spatial scales.
  • Extending the method to video or 3D volumes would require checking whether temporal or volumetric consistency improves OOD separation.

Load-bearing premise

OOD detection methods built for whole-image classification can be applied directly to individual pixels for segmentation without major redesign or large performance loss.

What would settle it

If pixel-wise application of standard OOD detectors produces segmentation accuracy substantially below that of fully supervised baselines on the same held-out surgical datasets, the central claim would be falsified.

Figures

Figures reproduced from arXiv: 2411.09553 by Jonathan Shapey, Junwen Wang, Oscar MacCormac, Tom Vercauteren, Zhonghao Wang.

Figure 1
Figure 1. Figure 1: Decision boundaries under different learning settings for a multi￾class problem with annotations from only positive classes. Coloured circle markers ( / / ) denote positively labelled samples for each class. Black question marks (?) indicate unused unlabelled training data, while coloured question marks (?/?/?) represent used unlabelled training data. Unlabelled data could be stemming from a positive class… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed OOD-SEG framework. During the training stage, only annotated pixels for the multiple positive classes are used to [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Graphical representation of the proposed OOD-focused two-level [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Graphical illustration of confusion matrix incorporating multi [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: demonstrates that our method effectively separates OOD pixels while maintaining classification accuracy for ID pixels [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: ROC curves comparing different OOD detection methods within our framework. The curve shows the average performance of 16 models over all two-level folds. Shaded area represents standard deviation across the folds. Experiments were conducted on the Heiporspectral dataset [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative result using all labelled classes during training, i.e. with N [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
read the original abstract

Despite significant advancements, segmentation based on deep neural networks in medical and surgical imaging faces several challenges, two of which we aim to address in this work. First, acquiring complete pixel-level segmentation labels for medical images is time-consuming and requires domain expertise. Second, typical segmentation pipelines cannot detect out-of-distribution (OOD) pixels, leaving them prone to spurious outputs during deployment. In this work, we propose a novel segmentation approach which broadly falls within the positive-unlabelled (PU) learning paradigm and exploits tools from OOD detection techniques. Our framework learns only from sparsely annotated pixels from multiple positive-only classes and does not use any annotation for the background class. These multi-class positive annotations naturally fall within the in-distribution (ID) set. Unlabelled pixels may contain positive classes but also negative ones, including what is typically referred to as \emph{background} in standard segmentation formulations. To the best of our knowledge, this work is the first to formulate multi-class segmentation with sparse positive-only annotations as a pixel-wise PU learning problem and to address it using OOD detection techniques. Here, we forgo the need for background annotation and consider these together with any other unseen classes as part of the OOD set. Our framework can integrate, at a pixel-level, any OOD detection approaches designed for classification tasks. To address the lack of existing OOD datasets and established evaluation metric for medical image segmentation, we propose a cross-validation strategy that treats held-out labelled classes as OOD. Extensive experiments on both multi-class hyperspectral and RGB surgical imaging datasets demonstrate the robustness and generalisation capability of our proposed framework.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes OOD-SEG, a framework that formulates multi-class segmentation from sparse positive-only annotations as a pixel-wise positive-unlabeled (PU) learning problem. It integrates out-of-distribution (OOD) detection techniques to treat unlabelled pixels (including background) as OOD without requiring background annotations, claims to be the first such approach, and introduces a cross-validation strategy that holds out labelled classes as OOD proxies. Experiments are reported on multi-class hyperspectral and RGB surgical imaging datasets to demonstrate robustness.

Significance. If the pixel-level adaptation of OOD detectors is shown to preserve their properties and the cross-validation proxy is validated, the work could meaningfully lower annotation costs in medical imaging while adding OOD awareness to segmentation models. The reliance on existing OOD methods rather than new inventions is a potential strength if the transfer is rigorously demonstrated.

major comments (2)
  1. [Abstract] Abstract: the central claim that the framework 'can integrate, at a pixel-level, any OOD detection approaches designed for classification tasks' is load-bearing for the contribution yet unsupported; no derivation, adaptation analysis, or ablation shows that whole-image OOD methods retain their guarantees or performance when applied to per-pixel feature maps rather than global embeddings.
  2. [Abstract] Abstract (cross-validation strategy): treating held-out labelled classes as OOD is presented as addressing the lack of OOD datasets, but this proxy is not shown to simulate deployment OOD (background, artifacts, unseen pathologies) whose pixel statistics differ from held-out ID classes drawn from the same sparse positive distribution; this directly affects the validity of the reported generalization claims.
minor comments (1)
  1. The abstract states 'extensive experiments' but provides no mention of specific metrics, baselines, or quantitative results; adding these details would improve clarity even if full results appear later in the manuscript.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript. We address each major comment below with clarifications and indicate planned revisions where appropriate to strengthen the presentation of our contributions.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the framework 'can integrate, at a pixel-level, any OOD detection approaches designed for classification tasks' is load-bearing for the contribution yet unsupported; no derivation, adaptation analysis, or ablation shows that whole-image OOD methods retain their guarantees or performance when applied to per-pixel feature maps rather than global embeddings.

    Authors: We acknowledge that the abstract states a broad integration claim without accompanying formal analysis in the provided text. The framework applies OOD detectors to per-pixel embeddings extracted from the segmentation network, treating each pixel independently, and the experiments demonstrate this with multiple methods. However, we agree a dedicated adaptation analysis is warranted to discuss transfer of properties from image-level to pixel-level application. We will revise the manuscript to include a new subsection deriving the pixel-wise formulation, analyzing retained detection characteristics, and expanding ablations on performance across OOD methods. revision: yes

  2. Referee: [Abstract] Abstract (cross-validation strategy): treating held-out labelled classes as OOD is presented as addressing the lack of OOD datasets, but this proxy is not shown to simulate deployment OOD (background, artifacts, unseen pathologies) whose pixel statistics differ from held-out ID classes drawn from the same sparse positive distribution; this directly affects the validity of the reported generalization claims.

    Authors: The cross-validation approach is introduced to enable quantitative evaluation in the absence of standard OOD benchmarks for this sparse annotation setting. Held-out classes serve as a controlled proxy for unseen positives. We recognize that such proxies drawn from the same distribution may not fully replicate the statistics of real deployment OOD elements like artifacts or pathologies. The manuscript reports additional results on actual unlabelled pixels (including background) in the surgical datasets to support robustness. We will add an explicit limitations discussion clarifying the proxy's scope and how the real-pixel evaluations bolster the generalization claims. revision: partial

Circularity Check

0 steps flagged

No circularity: framework integrates external OOD methods at pixel level without self-referential derivations or fitted inputs renamed as predictions

full rationale

The abstract and described approach formulate the task as pixel-wise PU learning and propose integrating existing OOD detection techniques designed for classification, with a cross-validation proxy for evaluation. No equations, derivations, or parameter-fitting steps are present that reduce by construction to the inputs (e.g., no self-definitional scaling, no fitted parameters called predictions, no load-bearing self-citations of uniqueness theorems). The central claim relies on external methods rather than internal circular construction, making the derivation self-contained against the provided text.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are described in the abstract; the framework is presented as an integration of existing OOD techniques into a PU learning formulation without introducing new postulated quantities.

pith-pipeline@v0.9.0 · 5843 in / 1173 out tokens · 26270 ms · 2026-05-23T17:42:41.606340+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

68 extracted references · 68 canonical work pages · 1 internal anchor

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

    work page

  2. [2]

    Concrete Problems in AI Safety

    author Amodei, D. , author Olah, C. , author Steinhardt, J. , author Christiano, P. , author Schulman, J. , author Man \'e , D. , year 2016 . title Concrete problems in ai safety . :10.48550/arXiv.1606.06565

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1606.06565 2016

  3. [3]

    , author Adler, T.J

    author Ayala, L. , author Adler, T.J. , author Seidlitz, S. , author Wirkert, S. , author Engels, C. , author Seitel, A. , author Sellner, J. , author Aksenov, A. , author Bodenbach, M. , author Bader, P. , author Baron, S. , author Vemuri, A. , author Wiesenfarth, M. , author Schreck, N. , author Mindroc, D. , author Tizabi, M. , author Pirmann, S. , aut...

    work page doi:10.1126/sciadv.add6778 2023

  4. [4]

    , author Horgan, C.C

    author Bahl, A. , author Horgan, C.C. , author Janatka, M. , author MacCormac, O.J. , author Noonan, P. , author Xie, Y. , author Qiu, J. , author Cavalcanti, N. , author F \"u rnstahl, P. , author Ebner, M. , author Bergholt, M.S. , author Shapey, J. , author Vercauteren, T. , year 2023 . title Synthetic white balancing for intra-operative hyperspectral ...

    work page doi:10.1117/1.jmi.10.4.046001 2023

  5. [5]

    , author Garcia-Peraza-Herrera , L

    author Barfoot, T. , author Garcia-Peraza-Herrera , L. , author Glocker, B. , author Vercauteren, T. , year 2024 . title Average calibration error: A differentiable loss for improved reliability in image segmentation . http://arxiv.org/abs/2403.06759

    work page arXiv 2024

  6. [6]

    Learning from positive and unlabeled data: a survey

    author Bekker, J. , author Davis, J. , year 2020 . title Learning from positive and unlabeled data: A survey . journal Machine Learning volume 109 , pages 719--760 . :10.1007/s10994-020-05877-5

    work page doi:10.1007/s10994-020-05877-5 2020

  7. [7]

    , year 2006

    author Bishop, C.M. , year 2006 . title Pattern Recognition and Machine Learning . Information Science and Statistics, address New York

    work page 2006

  8. [8]

    , author Nikdan, M

    author Bitarafan, A. , author Nikdan, M. , author Baghshah, M.S. , year 2021 . title 3d image segmentation with sparse annotation by self-training and internal registration . journal IEEE Journal of Biomedical and Health Informatics volume 25 , pages 2665--2672 . :10.1109/JBHI.2020.3038847

    work page doi:10.1109/jbhi.2020.3038847 2021

  9. [9]

    , author Kailkhura, B

    author Bulusu, S. , author Kailkhura, B. , author Li, B. , author Varshney, P.K. , author Song, D. , year 2020 . title Anomalous example detection in deep learning: A survey . journal IEEE Access volume 8 , pages 132330--132347 . :10.1109/ACCESS.2020.3010274

    work page doi:10.1109/access.2020.3010274 2020

  10. [10]

    , author Qi, L

    author Cai, H. , author Qi, L. , author Yu, Q. , author Shi, Y. , author Gao, Y. , year 2023 . title 3d medical image segmentation with sparse annotation via cross-teaching between 3d and 2d networks , in: booktitle Medical Image Computing and Computer Assisted Intervention -- MICCAI 2023: 26th International Conference , address Berlin, Heidelberg . pp. p...

    work page doi:10.1007/978-3-031-43898-1_59 2023

  11. [11]

    , author Chaitanya, K

    author Can, Y.B. , author Chaitanya, K. , author Mustafa, B. , author Koch, L.M. , author Konukoglu, E. , author Baumgartner, C.F. , year 2018 . title Learning to segment medical images with scribble-supervision alone , in: booktitle Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support , address Cham . pp. pages 23...

    work page doi:10.1007/978-3-030-00889-5_27 2018

  12. [12]

    , author Rinner, F.M

    author Carstens, M. , author Rinner, F.M. , author Bodenstedt, S. , author Jenke, A.C. , author Weitz, J. , author Distler, M. , author Speidel, S. , author Kolbinger, F.R. , year 2023 . title The dresden surgical anatomy dataset for abdominal organ segmentation in surgical data science . journal Scientific Data volume 10 , pages 3 . :10.1038/s41597-022-01719-2

    work page doi:10.1038/s41597-022-01719-2 2023

  13. [13]

    , author Dong, W

    author Deng, J. , author Dong, W. , author Socher, R. , author Li, L.J. , author Li, K. , author Fei-Fei , L. , year 2009 . title Imagenet: A large-scale hierarchical image database , in: booktitle IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pp. pages 248--255

    work page 2009

  14. [14]

    , author Joutard, S

    author Dorent, R. , author Joutard, S. , author Shapey, J. , author Kujawa, A. , author Modat, M. , author Ourselin, S. , author Vercauteren, T. , year 2021 . title Inter extreme points geodesics for end-to-end weakly supervised image segmentation , in: booktitle Medical Image Computing and Computer Assisted Intervention -- MICCAI 2021 , address Cham . pp...

    work page doi:10.1007/978-3-030-87196-3_57 2021

  15. [15]

    , author Ortega, S

    author Fabelo, H. , author Ortega, S. , author Ravi, D. , author Kiran, B.R. , author Sosa, C. , author Bulters, D. , author Callic \'o , G.M. , author Bulstrode, H. , author Szolna, A. , author Pi \ n eiro, J.F. , author Kabwama, S. , author Madro \ n al, D. , author Lazcano, R. , author J-O'Shanahan , A. , author Bisshopp, S. , author Hern \'a ndez, M. ...

    work page doi:10.1371/journal.pone.0193721 2018

  16. [16]

    , author Aertsen, M

    author Fidon, L. , author Aertsen, M. , author Emam, D. , author Mufti, N. , author Guffens, F. , author Deprest, T. , author Demaerel, P. , author David, A.L. , author Melbourne, A. , author Ourselin, S. , author Deprest, J. , author Vercauteren, T. , year 2021 . title Label-set loss functions for partial supervision: Application to fetal brain 3d mri pa...

    work page doi:10.1007/978-3-030-87196-3_60 2021

  17. [17]

    , author Ghahramani, Z

    author Gal, Y. , author Ghahramani, Z. , year 2016 . title Dropout as a bayesian approximation: Representing model uncertainty in deep learning , in: booktitle Proceedings of The 33rd International Conference on Machine Learning , pp. pages 1050--1059 . https://proceedings.mlr.press/v48/gal16.html

    work page 2016

  18. [18]

    , author Horgan, C

    author Garcia Peraza Herrera, L.C. , author Horgan, C. , author Ourselin, S. , author Ebner, M. , author Vercauteren, T. , year 2023 . title Hyperspectral image segmentation: A preliminary study on the oral and dental spectral image database (odsi-db) . journal Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization volume 11 ...

    work page doi:10.1080/21681163.2022.2160377 2023

  19. [19]

    , author Zikic, D

    author Glocker, B. , author Zikic, D. , author Konukoglu, E. , author Haynor, D.R. , author Criminisi, A. , year 2013 . title Vertebrae localization in pathological spine ct via dense classification from sparse annotations , in: booktitle Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2013 , address Berlin, Heidelberg . pp. pages 262...

    work page doi:10.1007/978-3-642-40763-5_33 2013

  20. [20]

    , author Gotkowski, K

    author Gonz \'a lez, C. , author Gotkowski, K. , author Fuchs, M. , author Bucher, A. , author Dadras, A. , author Fischbach, R. , author Kaltenborn, I.J. , author Mukhopadhyay, A. , year 2022 . title Distance-based detection of out-of-distribution silent failures for covid-19 lung lesion segmentation . journal Medical Image Analysis volume 82 , pages 102...

    work page doi:10.1016/j.media.2022.102596 2022

  21. [21]

    , author Pleiss, G

    author Guo, C. , author Pleiss, G. , author Sun, Y. , author Weinberger, K.Q. , year 2017 . title On calibration of modern neural networks , in: booktitle Proceedings of the 34th International Conference on Machine Learning , pp. pages 1321--1330 . https://proceedings.mlr.press/v70/guo17a.html

    work page 2017

  22. [22]

    , author He, R

    author Han, Z. , author He, R. , author Li, T. , author Wei, B. , author Wang, J. , author Yin, Y. , year 2021 . title Semi-supervised screening of covid-19 from positive and unlabeled data with constraint non-negative risk estimator , in: booktitle Information Processing in Medical Imaging , pp. pages 611--623 . :10.1007/978-3-030-78191-0_47

    work page doi:10.1007/978-3-030-78191-0_47 2021

  23. [23]

    , author Gimpel, K

    author Hendrycks, D. , author Gimpel, K. , year 2017 . title A baseline for detecting misclassified and out-of-distribution examples in neural networks , in: booktitle International Conference on Learning Representations , pp. pages 1--12 . https://openreview.net/forum?id=Hkg4TI9xl

    work page 2017

  24. [24]

    In: Proceedings of the IEEE/CVF Conference on Computer 25 Vision and Pattern Recognition, pp

    author Hsu, Y.C. , author Shen, Y. , author Jin, H. , author Kira, Z. , year 2020 . title Generalized odin: Detecting out-of-distribution image without learning from out-of-distribution data , in: booktitle 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pp. pages 10948--10957 . :10.1109/CVPR42600.2020.01096

    work page doi:10.1109/cvpr42600.2020.01096 2020

  25. [25]

    a lt, P. , author J \

    author Hyttinen, J. , author F \"a lt, P. , author J \"a sberg, H. , author Kullaa, A. , author Hauta-Kasari , M. , year 2020 . title Oral and dental spectral image database---odsi-db . journal Applied Sciences volume 10 , pages 7246 . :10.3390/app10207246

    work page doi:10.3390/app10207246 2020

  26. [26]

    , author Drozdzal, M

    author Jegou, S. , author Drozdzal, M. , author Vazquez, D. , author Romero, A. , author Bengio, Y. , year 2017 . title The one hundred layers tiramisu: Fully convolutional densenets for semantic segmentation , in: booktitle Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops , pp. pages 11--19 . https://openaccess.thec...

    work page 2017

  27. [27]

    , author Zhang, D

    author Jiang, X. , author Zhang, D. , author Li, X. , author Liu, K. , author Cheng, K.T. , author Yang, X. , year 2025 . title Labeled-to-unlabeled distribution alignment for partially-supervised multi-organ medical image segmentation . journal Medical Image Analysis volume 99 , pages 103333 . :10.1016/j.media.2024.103333

    work page doi:10.1016/j.media.2024.103333 2025

  28. [28]

    , author Gholipour, A

    author Karimi, D. , author Gholipour, A. , year 2023 . title Improving calibration and out-of-distribution detection in deep models for medical image segmentation . journal IEEE Transactions on Artificial Intelligence volume 4 , pages 383--397 . :10.1109/TAI.2022.3159510

    work page doi:10.1109/tai.2022.3159510 2023

  29. [29]

    , author Paheding, S

    author Khan, U. , author Paheding, S. , author Elkin, C.P. , author Devabhaktuni, V.K. , year 2021 . title Trends in deep learning for medical hyperspectral image analysis . journal IEEE Access volume 9 , pages 79534--79548 . :10.1109/ACCESS.2021.3068392

    work page doi:10.1109/access.2021.3068392 2021

  30. [30]

    , author Ba, J

    author Kingma, D.P. , author Ba, J. , year 2017 . title Adam: A method for stochastic optimization , in: booktitle 3rd International Conference on Learning Representations (ICLR), San Diego, 2015 , p. pages 8

    work page 2017

  31. [31]

    , author Yan, Z

    author Kuang, Z. , author Yan, Z. , author Yu, L. , year 2024 . title Weakly supervised learning for multi-class medical image segmentation via feature decomposition . journal Computers in Biology and Medicine volume 171 , pages 108228 . :10.1016/j.compbiomed.2024.108228

    work page doi:10.1016/j.compbiomed.2024.108228 2024

  32. [32]

    , author Pritzel, A

    author Lakshminarayanan, B. , author Pritzel, A. , author Blundell, C. , year 2017 . title Simple and scalable predictive uncertainty estimation using deep ensembles , in: booktitle Advances in Neural Information Processing Systems , pp. pages 1--12 . https://proceedings.neurips.cc/paper_files/paper/2017/hash/9ef2ed4b7fd2c810847ffa5fa85bce38-Abstract.html

    work page 2017

  33. [33]

    , author Forbes, F

    author Lambert, B. , author Forbes, F. , author Doyle, S. , author Dehaene, H. , author Dojat, M. , year 2024 . title Trustworthy clinical ai solutions: A unified review of uncertainty quantification in deep learning models for medical image analysis . journal Artificial Intelligence in Medicine volume 150 , pages 102830 . :10.1016/j.artmed.2024.102830

    work page doi:10.1016/j.artmed.2024.102830 2024

  34. [34]

    , author Lee, K

    author Lee, K. , author Lee, K. , author Lee, H. , author Shin, J. , year 2018 . title A simple unified framework for detecting out-of-distribution samples and adversarial attacks , in: booktitle Advances in Neural Information Processing Systems , pp. pages 1--12 . https://papers.nips.cc/paper_files/paper/2018/hash/abdeb6f575ac5c6676b747bca8d09cc2-Abstract.html

    work page 2018

  35. [35]

    , author Asad, M

    author Li, P. , author Asad, M. , author Horgan, C. , author MacCormac, O. , author Shapey, J. , author Vercauteren, T. , year 2023 . title Spatial gradient consistency for unsupervised learning of hyperspectral demosaicking: Application to surgical imaging . journal International Journal of Computer Assisted Radiology and Surgery volume 18 , pages 981--9...

    work page doi:10.1007/s11548-023-02865-7 2023

  36. [36]

    , author Ebner, M

    author Li, P. , author Ebner, M. , author Noonan, P. , author Horgan, C. , author Bahl, A. , author Ourselin, S. , author Shapey, J. , author Vercauteren, T. , year 2022 . title Deep learning approach for hyperspectral image demosaicking, spectral correction and high-resolution rgb reconstruction . journal Computer Methods in Biomechanics and Biomedical E...

    work page doi:10.1080/21681163.2021.1997646 2022

  37. [37]

    , author Li, L

    author Lian, S. , author Li, L. , author Luo, Z. , author Zhong, Z. , author Wang, B. , author Li, S. , year 2023 . title Learning multi-organ segmentation via partial- and mutual-prior from single-organ datasets . journal Biomedical Signal Processing and Control volume 80 , pages 104339 . :10.1016/j.bspc.2022.104339

    work page doi:10.1016/j.bspc.2022.104339 2023

  38. [38]

    , author Zhang, Y

    author Liang, G. , author Zhang, Y. , author Wang, X. , author Jacobs, N. , year 2020 . title Improved trainable calibration method for neural networks on medical imaging classification . http://arxiv.org/abs/2009.04057

    work page arXiv 2020

  39. [39]

    , author Li, Y

    author Liang, S. , author Li, Y. , author Srikant, R. , year 2018 . title Enhancing the reliability of out-of-distribution image detection in neural networks , in: booktitle International Conference on Learning Representations , pp. pages 1--12 . https://openreview.net/forum?id=H1VGkIxRZ

    work page 2018

  40. [40]

    , author Xu, Z

    author Liu, H. , author Xu, Z. , author Gao, R. , author Li, H. , author Wang, J. , author Chabin, G. , author Oguz, I. , author Grbic, S. , year 2024 . title Cosst: Multi-organ segmentation with partially labeled datasets using comprehensive supervisions and self-training . journal IEEE Transactions on Medical Imaging volume 43 , pages 1995--2009 . :10.1...

    work page doi:10.1109/tmi.2024.3354673 2024

  41. [41]

    , author Hinton, G

    author van der Maaten, L. , author Hinton, G. , year 2008 . title Visualizing data using t-sne . journal Journal of Machine Learning Research volume 9 , pages 2579--2605 . http://jmlr.org/papers/v9/vandermaaten08a.html

    work page 2008

  42. [42]

    , author Martinez-Vega , B

    author Mart \'i n-P \'e rez , A. , author Martinez-Vega , B. , author Villa, M. , author Leon, R. , author Martinez de Ternero , A. , author Fabelo, H. , author Ortega, S. , author Quevedo, E. , author Callico, G.M. , author Juarez, E. , author Sanz, C. , year 2024 . title Machine learning performance trends: A comparative study of independent hyperspectr...

    work page 2024

  43. [43]

    , author Wirkert, S.J

    author Moccia, S. , author Wirkert, S.J. , author Kenngott, H. , author Vemuri, A.S. , author Apitz, M. , author Mayer, B. , author De Momi, E. , author Mattos, L.S. , author Maier-Hein , L. , year 2018 . title Uncertainty-aware organ classification for surgical data science applications in laparoscopy . journal IEEE Transactions on Biomedical Engineering...

    work page doi:10.1109/tbme.2018.2813015 2018

  44. [44]

    , author Patel, V.M

    author Oza, P. , author Patel, V.M. , year 2019 . title One-class convolutional neural network . journal IEEE Signal Processing Letters volume 26 , pages 277--281 . :10.1109/LSP.2018.2889273

    work page doi:10.1109/lsp.2018.2889273 2019

  45. [45]

    , author Oza, P

    author Perera, P. , author Oza, P. , author Patel, V.M. , year 2021 . title One-class classification: A survey . :10.48550/arXiv.2101.03064

    work page doi:10.48550/arxiv.2101.03064 2021

  46. [46]

    , author Wu, P

    author Qu, H. , author Wu, P. , author Huang, Q. , author Yi, J. , author Riedlinger, G.M. , author De, S. , author Metaxas, D.N. , year 2019 . title Weakly supervised deep nuclei segmentation using points annotation in histopathology images , in: booktitle Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning , pp. pages 3...

    work page 2019

  47. [47]

    , author Fabelo, H

    author Rav \`i , D. , author Fabelo, H. , author Callic, G.M. , author Yang, G.Z. , year 2017 . title Manifold embedding and semantic segmentation for intraoperative guidance with hyperspectral brain imaging . journal IEEE Transactions on Medical Imaging volume 36 , pages 1845--1857 . :10.1109/TMI.2017.2695523

    work page doi:10.1109/tmi.2017.2695523 2017

  48. [48]

    9351, pp

    author Ronneberger, O. , author Fischer, P. , author Brox, T. , year 2015 . title U-net: Convolutional networks for biomedical image segmentation , in: booktitle Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2015 , address Cham . pp. pages 234--241 . :10.1007/978-3-319-24574-4_28

    work page doi:10.1007/978-3-319-24574-4_28 2015

  49. [49]

    , author Vandermeulen, R

    author Ruff, L. , author Vandermeulen, R. , author Goernitz, N. , author Deecke, L. , author Siddiqui, S.A. , author Binder, A. , author M \"u ller, E. , author Kloft, M. , year 2018 . title Deep one-class classification , in: booktitle Proceedings of the 35th International Conference on Machine Learning , pp. pages 4393--4402 . https://proceedings.mlr.pr...

    work page 2018

  50. [50]

    and Rehmsmeier, M

    author Saito, T. , author Rehmsmeier, M. , year 2015 . title The precision-recall plot is more informative than the roc plot when evaluating binary classifiers on imbalanced datasets . journal PLOS ONE volume 10 , pages e0118432 . :10.1371/journal.pone.0118432

    work page doi:10.1371/journal.pone.0118432 2015

  51. [51]

    O zdemir, B. , author Studier-Fischer , A. , author Kn \

    author Seidlitz, S. , author Sellner, J. , author Odenthal, J. , author \"O zdemir, B. , author Studier-Fischer , A. , author Kn \"o dler, S. , author Ayala, L. , author Adler, T.J. , author Kenngott, H.G. , author Tizabi, M. , author Wagner, M. , author Nickel, F. , author M \"u ller-Stich , B.P. , author Maier-Hein , L. , year 2022 . title Robust deep l...

    work page doi:10.1016/j.media.2022.102488 2022

  52. [52]

    , author Xie, Y

    author Shapey, J. , author Xie, Y. , author Nabavi, E. , author Bradford, R. , author Saeed, S.R. , author Ourselin, S. , author Vercauteren, T. , year 2019 . title Intraoperative multispectral and hyperspectral label-free imaging: A systematic review of in vivo clinical studies . journal Journal of Biophotonics volume 12 , pages e201800455 . :10.1002/jbi...

    work page doi:10.1002/jbio.201800455 2019

  53. [53]

    Medical Image Analysis 70, 101979 (May 2021)

    author Shi, G. , author Xiao, L. , author Chen, Y. , author Zhou, S.K. , year 2021 . title Marginal loss and exclusion loss for partially supervised multi-organ segmentation . journal Medical Image Analysis volume 70 , pages 101979 . :10.1016/j.media.2021.101979

    work page doi:10.1016/j.media.2021.101979 2021

  54. [54]

    , author Johnson, M

    author Shotton, J. , author Johnson, M. , author Cipolla, R. , year 2008 . title Semantic texton forests for image categorization and segmentation , in: booktitle 2008 IEEE Conference on Computer Vision and Pattern Recognition , pp. pages 1--8 . :10.1109/CVPR.2008.4587503

    work page doi:10.1109/cvpr.2008.4587503 2008

  55. [55]

    , author Seidlitz, S

    author Studier-Fischer , A. , author Seidlitz, S. , author Sellner, J. , author Bressan, M. , author \"O zdemir, B. , author Ayala, L. , author Odenthal, J. , author Knoedler, S. , author Kowalewski, K.F. , author Haney, C.M. , author Salg, G. , author Dietrich, M. , author Kenngott, H. , author Gockel, I. , author Hackert, T. , author M \"u ller-Stich , ...

    work page doi:10.1038/s41597-023-02315-8 2023

  56. [56]

    , author Hanbury, A

    author Taha, A.A. , author Hanbury, A. , year 2015 . title Metrics for evaluating 3d medical image segmentation: Analysis, selection, and tool . journal BMC Medical Imaging volume 15 , pages 29 . :10.1186/s12880-015-0068-x

    work page doi:10.1186/s12880-015-0068-x 2015

  57. [57]

    , author Jeyaseelan, L

    author Tajbakhsh, N. , author Jeyaseelan, L. , author Li, Q. , author Chiang, J.N. , author Wu, Z. , author Ding, X. , year 2020 . title Embracing imperfect datasets: A review of deep learning solutions for medical image segmentation . journal Medical Image Analysis volume 63 , pages 101693 . :10.1016/j.media.2020.101693

    work page doi:10.1016/j.media.2020.101693 2020

  58. [58]

    , author Le, Q

    author Tan, M. , author Le, Q. , year 2019 . title Efficientnet: Rethinking model scaling for convolutional neural networks , in: booktitle Proceedings of the 36th International Conference on Machine Learning , pp. pages 6105--6114 . https://proceedings.mlr.press/v97/tan19a.html

    work page 2019

  59. [59]

    , year 2024

    author Union, E. , year 2024 . title Artificial intelligence act (regulation (eu) 2024/1689) . https://artificialintelligenceact.eu/high-level-summary/

    work page 2024

  60. [60]

    , author Li, W

    author Wang, G. , author Li, W. , author Aertsen, M. , author Deprest, J. , author Ourselin, S. , author Vercauteren, T. , year 2019 a. title Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks . journal Neurocomputing volume 335 , pages 34--45 . :10.1016/j.neucom.2019.01.103

    work page doi:10.1016/j.neucom.2019.01.103 2019

  61. [61]

    , author Li, W

    author Wang, G. , author Li, W. , author Ourselin, S. , author Vercauteren, T. , year 2019 b. title Automatic brain tumor segmentation based on cascaded convolutional neural networks with uncertainty estimation . journal Frontiers in Computational Neuroscience volume 13 . :10.3389/fncom.2019.00056

    work page doi:10.3389/fncom.2019.00056 2019

  62. [62]

    , author Li, W

    author Wang, G. , author Li, W. , author Zuluaga, M.A. , author Pratt, R. , author Patel, P.A. , author Aertsen, M. , author Doel, T. , author David, A.L. , author Deprest, J. , author Ourselin, S. , author Vercauteren, T. , year 2018 . title Interactive medical image segmentation using deep learning with image-specific fine tuning . journal IEEE Transact...

    work page doi:10.1109/tmi.2018.2791721 2018

  63. [63]

    , author Zuluaga, M.A

    author Wang, G. , author Zuluaga, M.A. , author Li, W. , author Pratt, R. , author Patel, P.A. , author Aertsen, M. , author Doel, T. , author David, A.L. , author Deprest, J. , author Ourselin, S. , author Vercauteren, T. , year 2019 c. title Deepigeos: A deep interactive geodesic framework for medical image segmentation . journal IEEE Transactions on Pa...

    work page doi:10.1109/tpami.2018.2840695 2019

  64. [64]

    , author Xia, B

    author Wang, J. , author Xia, B. , year 2021 . title Bounding box tightness prior for weakly supervised image segmentation , in: booktitle Medical Image Computing and Computer Assisted Intervention -- MICCAI 2021 , address Cham . pp. pages 526--536 . :10.1007/978-3-030-87196-3_49

    work page doi:10.1007/978-3-030-87196-3_49 2021

  65. [65]

    , author Snoek, J

    author Wenzel, F. , author Snoek, J. , author Tran, D. , author Jenatton, R. , year 2020 . title Hyperparameter ensembles for robustness and uncertainty quantification , in: booktitle Advances in Neural Information Processing Systems , pp. pages 6514--6527 . https://proceedings.neurips.cc/paper/2020/hash/481fbfa59da2581098e841b7afc122f1-Abstract.html

    work page 2020

  66. [66]

    , author Zhu, J.Y

    author Xu, Y. , author Zhu, J.Y. , author Chang, E.I.C. , author Lai, M. , author Tu, Z. , year 2014 . title Weakly supervised histopathology cancer image segmentation and classification . journal Medical Image Analysis volume 18 , pages 591--604 . :10.1016/j.media.2014.01.010

    work page doi:10.1016/j.media.2014.01.010 2014

  67. [67]

    , author Zhou, K

    author Yang, J. , author Zhou, K. , author Li, Y. , author Liu, Z. , year 2024 . title Generalized out-of-distribution detection: A survey . journal International Journal of Computer Vision :10.1007/s11263-024-02117-4

    work page doi:10.1007/s11263-024-02117-4 2024

  68. [68]

    , author Deng, W

    author Zhu, Q. , author Deng, W. , author Zheng, Z. , author Zhong, Y. , author Guan, Q. , author Lin, W. , author Zhang, L. , author Li, D. , year 2022 . title A spectral-spatial-dependent global learning framework for insufficient and imbalanced hyperspectral image classification . journal IEEE Transactions on Cybernetics volume 52 , pages 11709--11723 ...

    work page doi:10.1109/tcyb.2021.3070577 2022