Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

From Deterministic to Probabilistic: A Novel Perspective on Domain Generalization for Medical Image Segmentation

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that distribution-level contrastive learning on class mean and covariance, combined with wavelet-based structure preservation, achieves the best domain-generalized medical segmentation on two benchmarks.

desk verdict Plausible empirical recipe for medical-image domain generalization, but the core probabilistic-contrastive derivation is missing and, as written, Eq. (5) does not follow from Eq. (4). read the letter →

arxiv 2412.05572 v1 pith:WIGBPHF7 submitted 2024-12-07 cs.CV

Yuheng Xu , Taiping Zhang This is my paper · ORCID
classification cs.CV
keywords domaingeneralizationprobabilisticrepresentationlearningcontrastivecovariancemodelingdiscretewavelettransformmedicalimagesegmentationuncertaintycross-domain
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that domain generalization in medical image segmentation is better served by enriching feature representations than by trying to erase domain differences. Its framework models each semantic class as a distribution, capturing class-conditional mean and covariance, and aligns those distributions across augmentations with a contrastive loss. It then uses discrete wavelet transforms to keep high-frequency structure intact when style-transfer augmentations distort it. On the Fundus and Prostate benchmarks the combined method reports 90.05% and 89.94% average Dice with ASD 9.18 and 0.97, exceeding prior methods. If the claim holds, segmentation models could become more reliable on images from unseen scanners and protocols without explicit domain alignment.

What carries the argument

The load-bearing object is the probabilistic representation contrastive loss: per-class online statistics, mean $\mu^k$ (Eq. 2) and covariance $\Sigma^k$ (Eq. 3), turn pixel features into distributional prototypes, and Eq. (5) replaces the pixel-wise contrastive loss with the closed-form exponent $q^\top \mu/\tau + q^\top \Sigma q/(2\tau^2)$ for positive and negative classes. The second machine is wavelet-enhanced structural preservation: a discrete wavelet transform splits features into LL (style) and LH/HL/HH (structure) sub-bands, and the source image's high-frequency sub-bands are added to the style-transformed features under a learned sigmoid gate before inverse DWT reconstruction (Eqs. 7-9). Together they align distributions across domains and keep structural detail intact under style transfer.

What would settle it

On a trained model, compute the empirical pixel-wise contrastive loss (Eq. 4) and the closed-form loss (Eq. 5) over the same feature maps and class statistics; if the two values or their gradients disagree beyond small numerical noise as M and N grow, the claimed derivation is false and the method's mechanism is not the probabilistic alignment it describes.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a semantic segmentation model can generalize across domains by enriching its representation instead of aligning domains away: each semantic class is summarized by an online mean and covariance of pixel features, the contrastive loss is written directly on those statistics, and a wavelet transform preserves the high-frequency structure that style augmentation distorts. The paper reports that this combination reaches 90.05% average Dice with ASD 9.18 on the Fundus benchmark and 89.94% Dice with ASD 0.97 on the Prostate benchmark, outperforming the previous best method WT-PSE by 0.85 and 1.02 Dice points respectively. The reported ablation shows covariance modeling and wavelet enhancement each add a clear increment. The paper also states that Eq. (5) is a derived closed form of the pixel-wise contrastive loss, with the derivation deferred to an appendix that is absent from this version.

Load-bearing premise

The load-bearing premise is that per-class pixel features are Gaussian enough that the closed-form contrastive exponent in Eq. (5), $q^\top \mu/\tau + q^\top \Sigma q/(2\tau^2)$, is a faithful replacement for the pixel-wise loss in Eq. (4); the paper defers the proof to an appendix that is not present in this version.

Editorial extensions

If this is right

  • On the Fundus benchmark, the full method exceeds the previous best (WT-PSE) by 0.85% average Dice and lowers ASD from 10.12 to 9.18 voxels.
  • On the Prostate benchmark, it exceeds WT-PSE by 1.02% Dice and lowers ASD from 1.41 to 0.97 voxels.
  • Ablations on Fundus attribute about 2.09 Dice points to adding covariance to mean-only probabilistic prototypes, and about 1.27 Dice points to the wavelet module on top of PRL with covariance.
  • The closed-form contrastive loss avoids enumerating large sets of pixel pairs, so distribution-level alignment can be applied at feature-map scale without the memory cost of pixel-wise contrastive learning.
  • Because the framework does not attempt to remove domain shift explicitly, it may keep working in settings where domain alignment cannot be fully achieved.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct numerical comparison of Eq. (4) and Eq. (5) on real feature maps would test whether the closed-form loss is truly equivalent; the paper leaves this unverified because the derivation appendix is absent.
  • The two modules should transfer to other dense prediction tasks that use style augmentation, such as driving-scene segmentation, since no domain labels are required.
  • The covariance term may matter most for classes with high intra-class variation; ablating per-class covariance separately, for example only for boundary classes, would test that.
  • The wavelet module's contribution is entangled with the TriD augmentation; swapping TriD for another augmentation would reveal whether WESP is augmentation-specific.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a domain-generalization framework for medical image segmentation that combines probabilistic representation contrastive learning with wavelet-enhanced structural preservation. The method models class-conditional feature statistics (mean and covariance), replaces pixel-wise contrastive loss with a closed-form Gaussian form in Eq. (5), and uses discrete wavelet transforms to inject source-domain high-frequency information into style-augmented features. Experiments on the Fundus and Prostate datasets under leave-one-domain-out evaluation report average Dice of 90.05% and 89.94%, respectively, exceeding the previous best method WT-PSE by 0.85 and 1.02 percentage points. An ablation study on Fundus attributes the gains to covariance modeling and to the wavelet-based structural preservation module.

Significance. If the reported results hold, the paper offers a useful and relatively simple alternative to explicit domain alignment, with a well-motivated frequency-domain component for preserving structural details under style augmentation. Strengths include the leave-one-domain-out protocol on two established medical-image benchmarks, three-run averaging, and an ablation study that isolates the contributions of the main components. The main limitation is that the core derivation of Eq. (5) is deferred to an appendix that is absent, and the empirical claims are reported without variance estimates, so the significance of the stated superiority over the prior state of the art is not yet established.

major comments (3)
  1. [III-A2, Eq. (5)] The text states that Eq. (5) is a 'special form' of the pixel-wise contrastive loss Eq. (4) and defers the proof to an appendix, but no appendix is present in this version. As written, Eq. (5) is not a special case of Eq. (4): Eq. (4) averages log-ratios over M positive samples, whereas Eq. (5) applies the Gaussian moment-generating function inside the log, and Jensen's inequality gives -E log R >= -log E R with equality only in degenerate cases. Please provide the full derivation, or alternatively state clearly that Eq. (5) is an approximation and validate it numerically (e.g., compare the two losses on a held-out batch) and in ablations.
  2. [IV-B, Tables I and II] The paper reports that each experiment was repeated three times but gives only averaged Dice and ASD values without standard deviations, confidence intervals, or significance tests. The headline improvements over WT-PSE are 0.85 Dice on Fundus (90.05 vs 89.20) and 1.02 Dice on Prostate (89.94 vs 88.92); without run-to-run variance estimates these differences are not shown to be statistically meaningful. Please report per-run results or error bars and, ideally, a paired significance test.
  3. [III-A1, Eq. (1)-(3)] The estimation of the class statistics used in Eq. (5) is underspecified. Eq. (1) defines mu'_s for source features, but the text then says that online statistics are updated using F_t from synthetic images; it is unclear whether mu_+, Sigma_+, mu_k-, and Sigma_k- in Eq. (5) are estimated from source features, augmented features, or a combination. Please specify exactly which features contribute to each statistic and whether the same statistics are used for all queries in Eq. (6).
minor comments (5)
  1. [IV-C] The text references 'Fig. 2' for visual comparison results, but the figure showing qualitative segmentation comparisons is Fig. 3; please correct the cross-reference.
  2. [III-A2, IV-D] The phrase 'Combining PRL with contrastive learning' is confusing because PRL already includes the contrastive loss; please reword to describe which components are actually being combined (e.g., covariance modeling and WESP) or define the rows of Table III more explicitly.
  3. [III-A1] The per-image covariance matrix Sigma'_j is used in Eq. (3) but never defined with a formula; adding its explicit definition would improve reproducibility.
  4. [IV-D] The sentence 'The baseline model uses TriD for data augmentation without incorporating PRL or WESP, achieving an average Dice score of 85.63' is missing a period after the number in the manuscript text; please fix the punctuation.
  5. [III-B] The wavelet sub-band notation and the enhancement formula in Eq. (8) would benefit from explicit tensor shapes and a precise description of how the sigmoid gate is applied, as this affects implementation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: held-out benchmarks, no target-domain fitting, no load-bearing self-citation chain.

full rationale

No circular step is present. The evaluation uses the standard leave-one-domain-out protocol: the model is trained on source domains plus style-augmented source features, and tested on a held-out target domain, so the reported Dice and ASD values are genuine held-out measurements rather than quantities fitted to the target. The probabilistic contrastive loss in Eq. (5) uses online class-mean and covariance statistics, which are computed from the model's own source/augmented features; this is standard self-supervised estimation, not a parameter fitted to the result it is then used to predict. The paper does not invoke a uniqueness theorem from the authors, and no load-bearing claim is justified only by a self-citation chain; the cited style-transfer and comparison methods are external prior work. The missing appendix derivation connecting Eq. (4) to Eq. (5) is a verifiability and correctness concern, but it is not circularity: even if the Gaussian-exponential reduction is invalid, the implemented loss would be a heuristic mean/covariance alignment, not a prediction that equals its input by construction. Accordingly, the central claim retains independent empirical content and the paper is scored 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central mechanism rests on two fitted or unspecified constants (contrastive temperature tau and DWT configuration) and three assumptions: Gaussian class-conditional feature distributions (Eq. 5), the wavelet style/structure separation, and the validity of online statistics updated from augmented features only. No invented entities are introduced.

free parameters (2)
  • Contrastive temperature tau = not reported
    Appears in Eqs. (4) and (5); controls the sharpness of the distribution-level contrastive loss and is set by hand, but no value or schedule is given.
  • DWT wavelet basis and decomposition level = not reported
    WESP depends on a specific discrete wavelet transform (family and level) to separate LL from high-frequency sub-bands; the implementation does not state these choices.
assumptions (3)
  • ad hoc to paper Per-class pixel features follow an approximately Gaussian distribution, so the pixel-wise contrastive loss can be collapsed to terms q^T mu/tau + q^T Sigma q/(2 tau^2).
    This is the mathematical content of the 'special form' of contrastive loss in Eq. (5). It is asserted without proof and without empirical validation of Gaussianity.
  • domain assumption In the discrete wavelet decomposition, the LL sub-band primarily contains style information while LH/HL/HH sub-bands primarily contain structural information.
    Section III.B states this to justify enhancing high-frequency sub-bands of augmented features; it is a common but not universal property of wavelet decompositions and is not verified here.
  • domain assumption Global class statistics aggregated online across source and augmented images are accurate estimates of class-conditional feature distributions; the recursive updates (2) and (3) are used with no correction for the fact that only augmented features update the statistics.
    Section III.A.1 says statistics are updated using synthetic (augmented) features F_t during training, which may bias prototypes toward augmented styles; the paper does not analyze this.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Deterministic to Probabilistic: A Novel Perspective on Domain Generalization for Medical Image Segmentation." pith.science (2026). https://pith.science/paper/WIGBPHF7

@misc{pith2026241205572,
  author       = {Pith},
  title        = {Pith review of: From Deterministic to Probabilistic: A Novel Perspective on Domain Generalization for Medical Image Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WIGBPHF7}},
  note         = {Machine review of arXiv:2412.05572}
}
read the original abstract

Traditional domain generalization methods often rely on domain alignment to reduce inter-domain distribution differences and learn domain-invariant representations. However, domain shifts are inherently difficult to eliminate, which limits model generalization. To address this, we propose an innovative framework that enhances data representation quality through probabilistic modeling and contrastive learning, reducing dependence on domain alignment and improving robustness under domain variations. Specifically, we combine deterministic features with uncertainty modeling to capture comprehensive feature distributions. Contrastive learning enforces distribution-level alignment by aligning the mean and covariance of feature distributions, enabling the model to dynamically adapt to domain variations and mitigate distribution shifts. Additionally, we design a frequency-domain-based structural enhancement strategy using discrete wavelet transforms to preserve critical structural details and reduce visual distortions caused by style variations. Experimental results demonstrate that the proposed framework significantly improves segmentation performance, providing a robust solution to domain generalization challenges in medical image segmentation.

Figures

Figures reproduced from arXiv: 2412.05572 by the authors.

Figure 1
Figure 1. Traditional domain alignment methods struggle to completely elim [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall architecture of our method. (a)Probabilistic Representation Learning: We construct probabilistic representation learning by leveraging [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visual comparison for Fundus and Prostate segmentation task. The red contours indicate the boundaries of ground truths while the green and blue contours are predictions [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. DistMedVL: Distributional Vision-Language Alignment for Uncertainty-Aware Medical Image Segmentation

    cs.CV 2026-08 conditional novelty 6.0 of 10

    A probabilistic vision-language adapter that models text tokens as Gaussian distributions and uses Mahalanobis distance for uncertainty-weighted alignment, improving medical image segmentation under data scarcity and ...

Reference graph

Works this paper leans on

23 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    Domain generalization via invariant feature representation,

    Krikamol Muandet, David Balduzzi, and Bernhard Sch ¨olkopf, “Domain generalization via invariant feature representation,” in International conference on machine learning . PMLR, 2013, pp. 10–18

    work page 2013

  2. [2]

    Domain generalization with adversarial feature learning,

    Haoliang Li, Sinno Jialin Pan, Shiqi Wang, and Alex C Kot, “Domain generalization with adversarial feature learning,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 5400–5409

    work page 2018

  3. [3]

    Deep domain generalization via conditional invariant adversarial networks,

    Ya Li, Xinmei Tian, Mingming Gong, Yajing Liu, Tongliang Liu, Kun Zhang, and Dacheng Tao, “Deep domain generalization via conditional invariant adversarial networks,” in Proceedings of the European confer- ence on computer vision (ECCV) , 2018, pp. 624–639

    work page 2018

  4. [4]

    Generalizable medical image segmentation via random amplitude mixup and domain-specific image restoration,

    Ziqi Zhou, Lei Qi, and Yinghuan Shi, “Generalizable medical image segmentation via random amplitude mixup and domain-specific image restoration,” in European Conference on Computer Vision . Springer, 2022, pp. 420–436

    work page 2022

  5. [5]

    Generalizing to unseen domains: A survey on domain generalization,

    Jindong Wang, Cuiling Lan, Chang Liu, Yidong Ouyang, Tao Qin, Wang Lu, Yiqiang Chen, Wenjun Zeng, and S Yu Philip, “Generalizing to unseen domains: A survey on domain generalization,” IEEE transactions on knowledge and data engineering, vol. 35, no. 8, pp. 8052–8072, 2022

    work page 2022

  6. [6]

    Domain generalization: A survey,

    Kaiyang Zhou, Ziwei Liu, Yu Qiao, Tao Xiang, and Chen Change Loy, “Domain generalization: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 4, pp. 4396–4415, 2022

    work page 2022

  7. [7]

    Treasure in distribution: a domain randomization based multi- source domain generalization for 2d medical image segmentation,

    Ziyang Chen, Yongsheng Pan, Yiwen Ye, Hengfei Cui, and Yong Xia, “Treasure in distribution: a domain randomization based multi- source domain generalization for 2d medical image segmentation,” in International Conference on Medical Image Computing and Computer- Assisted Intervention. Springer, 2023, pp. 89–99

    work page 2023

  8. [8]

    Seit: Structural enhancement for unsupervised image translation in frequency domain,

    Zhifeng Zhu, Yaochen Li, Yifan Li, Jinhuo Yang, Peijun Chen, and Yuehu Liu, “Seit: Structural enhancement for unsupervised image translation in frequency domain,” inProceedings of the AAAI Conference on Artificial Intelligence , 2024, vol. 38, pp. 7820–7827

    work page 2024

Show all 23 references
  1. [9]

    Mixstyle neural networks for domain generalization and adaptation,

    Kaiyang Zhou, Yongxin Yang, Yu Qiao, and Tao Xiang, “Mixstyle neural networks for domain generalization and adaptation,” International Journal of Computer Vision , vol. 132, no. 3, pp. 822–836, 2024

    2024

  2. [10]

    Learning to learn single domain generalization,

    Fengchun Qiao, Long Zhao, and Xi Peng, “Learning to learn single domain generalization,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 12556–12565

    2020

  3. [11]

    Selfreg: Self-supervised contrastive regularization for domain gen- eralization,

    Daehee Kim, Youngjun Yoo, Seunghyun Park, Jinkyu Kim, and Jaekoo Lee, “Selfreg: Self-supervised contrastive regularization for domain gen- eralization,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 9619–9628

    2021

  4. [12]

    Sepico: Semantic-guided pixel contrast for domain adaptive semantic segmentation,

    Binhui Xie, Shuang Li, Mingjia Li, Chi Harold Liu, Gao Huang, and Guoren Wang, “Sepico: Semantic-guided pixel contrast for domain adaptive semantic segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 7, pp. 9004–9021, 2023

    2023

  5. [13]

    Learning to generalize: Meta-learning for domain generalization,

    Da Li, Yongxin Yang, Yi-Zhe Song, and Timothy Hospedales, “Learning to generalize: Meta-learning for domain generalization,” in Proceedings of the AAAI conference on artificial intelligence , 2018, vol. 32

    2018

  6. [14]

    V-net: Fully convolutional neural networks for volumetric medical image segmentation,

    Fausto Milletari, Nassir Navab, and Seyed-Ahmad Ahmadi, “V-net: Fully convolutional neural networks for volumetric medical image segmentation,” in 2016 fourth international conference on 3D vision (3DV). Ieee, 2016, pp. 565–571

    2016

  7. [15]

    Kevin P Murphy, Machine learning: a probabilistic perspective , MIT press, 2012

    2012

  8. [16]

    Cddsa: Contrastive domain disentanglement and style augmentation for generalizable medical image segmentation,

    Ran Gu, Guotai Wang, Jiangshan Lu, Jingyang Zhang, Wenhui Lei, Yinan Chen, Wenjun Liao, Shichuan Zhang, Kang Li, Dimitris N Metaxas, et al., “Cddsa: Contrastive domain disentanglement and style augmentation for generalizable medical image segmentation,” Medical Image Analysis,...

    2023

  9. [17]

    Learning robust shape regularization for generalizable medical image segmentation,

    Kecheng Chen, Tiexin Qin, Victor Ho-Fun Lee, Hong Yan, and Haoliang Li, “Learning robust shape regularization for generalizable medical image segmentation,” IEEE Transactions on Medical Imaging , 2024

    2024

  10. [18]

    Domain generalization with correlated style uncertainty,

    Zheyuan Zhang, Bin Wang, Debesh Jha, Ugur Demir, and Ulas Bagci, “Domain generalization with correlated style uncertainty,” in Proceed- ings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2024, pp. 2000–2009

    2024

  11. [19]

    From denoising training to test-time adaptation: Enhancing domain generalization for medical image segmentation,

    Ruxue Wen, Hangjie Yuan, Dong Ni, Wenbo Xiao, and Yaoyao Wu, “From denoising training to test-time adaptation: Enhancing domain generalization for medical image segmentation,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2024, pp. 464–474

    2024

  12. [20]

    Dofe: Domain-oriented feature embedding for generalizable fundus image segmentation on unseen datasets,

    Shujun Wang, Lequan Yu, Kang Li, Xin Yang, Chi-Wing Fu, and Pheng- Ann Heng, “Dofe: Domain-oriented feature embedding for generalizable fundus image segmentation on unseen datasets,” IEEE Transactions on Medical Imaging, vol. 39, no. 12, pp. 4237–4248, 2020

    2020

  13. [21]

    Shape-aware meta-learning for generalizing prostate mri segmentation to unseen domains,

    Quande Liu, Qi Dou, and Pheng-Ann Heng, “Shape-aware meta-learning for generalizing prostate mri segmentation to unseen domains,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2020: 23rd International Conference, Lima, Peru, October 4–8, 2020, Proceeding...

    2020

  14. [22]

    Domain specific convolution and high frequency reconstruction based unsupervised domain adap- tation for medical image segmentation,

    Shishuai Hu, Zehui Liao, and Yong Xia, “Domain specific convolution and high frequency reconstruction based unsupervised domain adap- tation for medical image segmentation,” in International Conference on Medical Image Computing and Computer-Assisted Intervention . Springer, 2...

    2022

  15. [23]

    Deep residual learning for image recognition,

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770– 778

    2016

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.