Pith. sign in

REVIEW 4 major objections 5 minor 64 references

Neighbor Does Matter: Density-Aware Contrastive Learning for Medical Semi-supervised Segmentation

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read The paper claims that feature density alone can provide supervisory signal for semi-supervised multi-organ segmentation.

desk verdict Derivative of Hunting Sparsity, missing the key baseline; plausible refinements but 'SOTA' claim unsubstantiated as written. read the letter →

arxiv 2412.19871 v1 pith:WNWUR52H submitted 2024-12-27 cs.CV cs.LG

classification cs.CVcs.LG
keywords semi-supervisedsegmentationmedicalimagemulti-organcontrastivelearningfeaturedensitydensity-awareneighborgraphsfeature-spacegeometrypseudo-labeling
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

The paper argues that semi-supervised multi-organ segmentation can be improved by using the geometry of the feature space as an additional training signal. Its proposal, Density-Aware Contrastive Learning (DACL), estimates how dense each class feature is among its neighbors, treats low-density features as under-trained anchors, and pulls them toward a cluster center built from high-density features of the same class. This is meant to shrink the volume of each class cluster and increase intra-class compactness, which should help a co-teaching segmentation model handle unlabeled data with scarce labels. The authors report the best Dice scores on the ACDC and Synapse benchmarks among the methods compared, with the largest gains in the most label-scarce settings. If the claim holds, semi-supervised segmentation can exploit neighborhood information that pseudo-labeling and consistency regularization alone leave unused.

What carries the argument

The central object is a density-aware neighbor graph built from feature embeddings pooled per class by masked average pooling. Density of a feature is the average cosine similarity to its $k$ nearest neighbors in a memory bank and current batch, averaged over several $k$ values so that local and slightly larger neighborhoods both contribute. The graph yields two ranked sets: the lowest-density features become anchors, and the highest-density features from both the batch and the memory bank form a class center. A positiveness score between each anchor and the center, computed as a softmax-normalized cosine similarity with a scaling factor, is inserted into a SupCon-style contrastive loss, so anchors are pulled toward the center with strength that depends on how far they are. Claim 1 formalizes the optimal similarity in that loss as the normalized positiveness score.

What would settle it

Shuffle the density scores before anchor selection while keeping every other component fixed: if the Dice score does not drop, locating sparse regions is not what drives the gain. Separately, corrupt a small fraction of pseudo-labels during training; if the density-aware pull visibly amplifies those errors, the assumption that the density estimate stays reliable under label noise is refuted.

Watch

Extended reading notes

Core claim

On its own terms, the paper claims that the supervisory information needed for semi-supervised segmentation is already present in the geometry of the feature space. Within each class, features that have low average cosine similarity to their nearest neighbors mark sparse, under-trained regions; features with high density approximate the class center. DACL samples the low-density features as anchors, draws positive keys from high-density features in both the current batch and a memory bank, and uses a soft density-guided contrastive loss to pull anchors toward the approximated center. The paper states Claim 1, that the optimal similarity between an anchor and its class center equals the positiveness score normalized across anchors, giving a formal reason to weight low-density anchors by their distance to the center. Empirically, the method reports a new best result: 90.91 Dice on ACDC with 10% labels and 41.32 Dice on Synapse with 20% labels, along with better Jaccard and surface-distance metrics.

Load-bearing premise

The method assumes that the cosine-similarity density of a feature among its nearest neighbors, computed inside class masks that come partly from pseudo-labels, reliably identifies under-trained sparse features rather than outliers or noise.

Editorial extensions

If this is right

  • Training with DACL should yield denser, more separable class clusters in embedding space, as measured by Silhouette, Davies-Bouldin, and V-Measure in the paper.
  • On ACDC with 10% labels, DACL reaches 90.91 Dice and 0.38 average surface distance, exceeding the compared semi-supervised methods; on Synapse with 20% labels, it reaches 41.32 Dice.
  • The multi-scale density estimator and memory bank are necessary: ablations show removing them reduces Dice by roughly 0.8 to 1.1 points each.
  • Because DACL is a plug-in contrastive regularizer on a co-training baseline, it can be combined with other pseudo-labeling and consistency losses rather than replacing them.

Reading between the lines

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

  • Inference: The density-ranked anchor selection could double as an indicator of pseudo-label unreliability; a natural test is whether filtering out the lowest-density anchors instead of pulling them helps when label noise is high.
  • Inference: The same density-aware neighbor graph could be applied to the supervised branch, using ground-truth masks instead of pseudo-labels, which might tighten clusters further in the fully supervised regime.
  • Inference: The approach's dependence on class-pooled prototypes suggests it should transfer to other dense prediction tasks with long-tailed classes, such as instance or panoptic segmentation, though the paper does not test this.
  • Inference: The reported gains are on two relatively small benchmark datasets; a stronger test would evaluate on a larger multi-organ dataset with more annotation imbalance.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes Density-Aware Contrastive Learning (DACL) for semi-supervised multi-organ segmentation. The method builds on a co-teaching framework, estimates per-class feature density via multi-scale k-nearest-neighbor graphs over a memory bank, selects low-density features as anchors and high-density features as positive keys approximating cluster centers, and applies a soft density-guided contrastive loss. The reported experiments cover ACDC and Synapse under 5%/10% and 10%/20% labeled settings, with state-of-the-art claims, ablations, and hyperparameter sensitivity studies.

Significance. If fully supported, the paper would be a useful incremental contribution to semi-supervised medical image segmentation, extending density-guided contrastive learning with multi-scale density estimation, a memory bank, and soft positiveness weighting. The geometric intuition is reasonable, and the reported gains over several recent baselines are encouraging. However, the empirical case as written is incomplete: the closest prior method with the same core mechanism, Hunting Sparsity (Wang et al. 2023), is cited but never compared against, and the manuscript contains several unresolved formal and reproducibility gaps (undefined variables, missing appendix, and an unproved Claim 1). These issues must be addressed before the central state-of-the-art claim can be accepted.

major comments (4)
  1. [Table 1; Related Work] The closest prior work, Hunting Sparsity (Wang et al. 2023), which already uses density-aware neighbor graphs, low-density anchors, high-density positive keys approximating cluster centers, and a contrastive pull, is cited in the introduction but is absent from Table 1 and from the ablation study. Because DACL's mechanism is largely inherited from that method, the claim that DACL 'surpasses existing methods under all settings' is not substantiated without a direct comparison. Please add Hunting Sparsity to the ACDC and Synapse comparisons, and include an ablation that isolates DACL's own contributions (soft positiveness weights, multi-scale density, memory bank) from the shared density-guided contrastive component.
  2. [Dataset and Evaluation Metrics; Experiments and Results] The paper states in the dataset section that it evaluates on 'four public datasets' but then describes and reports results only for ACDC and Synapse; the abstract likewise mentions only the Multi-Organ Segmentation Challenge dataset. This inconsistency makes the empirical scope unclear and directly affects the strength of the claimed state-of-the-art. Please either correct the statement to two datasets or complete the evaluation on the missing datasets.
  3. [Methodology; Eq. 1; Eq. 7; Eq. 9] The main text repeatedly refers to an appendix for definitions of the supervised loss L_sup and the cross-supervised loss L_cross, but no appendix is present. In addition, Eq. (7) introduces the undefined quantity theta_{global/local}, and Eq. (9) uses an undefined scaling factor gamma_i. These omissions prevent reproduction of the method. Please provide the missing loss definitions (in the main text or a complete appendix) and define all symbols in Eqs. (7) and (9).
  4. [Claim 1] Claim 1 asserts that the optimal similarity measure s_i^* equals w_i / sum_k w_k, but s_i^* is not defined anywhere, and no proof is provided in the manuscript or in the referenced (absent) appendix. As written, this formal claim is unsupported and should either be proved with a clear statement of the optimization problem and assumptions, or removed from the paper.
minor comments (5)
  1. [Abstract; Section 'Dataset and Evaluation Metrics'] The abstract says the method is evaluated only on the Multi-Organ Segmentation Challenge dataset, while the main text reports ACDC and Synapse; please harmonize these statements.
  2. [Table 1] The layout of Table 1 is confusing: the 'CPS' row appears to cover two different scan settings, and the alignment of methods with their metrics is hard to follow. Please reformat the table so each method and each labeled/unlabeled split is clearly associated with its results.
  3. [Eq. 9] The notation l1(.) and l2(.) for 'parameter-free identity mapping layers' is unnecessary and confusing; if the embeddings are normalized, this should be stated once, and the transpose in the product should be made explicit.
  4. [Fig. 5] The caption of Fig. 5 labels both the threshold plot and the temperature plot as '(c)'; the second should be '(d)'.
  5. [Fig. 1] The text 'Davies-Boulding' should be 'Davies-Bouldin', and the metric abbreviation is usually DB, not D-B.

Circularity Check

1 steps flagged · score 2.0 of 10

No significant circularity: the reported segmentation gains rest on held-out ACDC/Synapse benchmarks, while the only circular step is a self-referential optimality claim (Claim 1) that does not affect the experimental conclusions.

  1. self definitional [Soft Density-Guided Contrastive Learning, Claim 1 (after Eq. 10)]
    "Claim 1. Assume we train models using the proposed optimization method, pn,+center and Mn are n-th class high-density positive keys and low-density anchor set, respectively. The optimal value of similarity measure s∗i can be expressed as wiPNqk=1 wk, where wi is the corresponding positiveness score for the prototype pair (pn,+center, mi ∈ Mn) in Eq. 10."

    Eq. 9 defines wi as a softmax (normalized exponential) of exactly the same anchor-to-center inner product that Claim 1 treats as the unknown similarity to be optimized: wi = (1/γi) softmax[l1(mn_i)^T l2(pcenter)]. A softmax value already is exp(s_i)/Σ exp(s_j), so dividing it by Σ wk returns the same softmax probability. Claim 1 therefore restates the definition of wi instead of deriving an optimum from Eq. 10; it is an identity, not an independent result. The held-out Dice/JC/ASD results do not depend on this claim.

full rationale

The central empirical claim—DACL surpasses prior methods on ACDC and Synapse—is supported by held-out external benchmarks and standard baselines, so it is not forced by construction. The density estimates, positive keys, and positiveness weights are computed from the same feature embeddings that the contrastive loss rearranges, but that is an optimization loop rather than a disguised prediction; the reported Dice/JC/ASD are measured on held-out data and are independent of that loop. The one genuinely self-referential item is Claim 1, which is tautological and not used to produce the experimental numbers. Self-citations in the introduction and related work (e.g., Tang et al. 2024b) are contextual and not load-bearing. The absence of Hunting Sparsity (Wang et al. 2023)—the closest density-guided contrastive predecessor—from Table 1 is a missing-control concern for the SOTA claim, but that is an experimental-design issue, not circularity.

Assumptions & free parameters 7 free parameters · 3 assumptions · 0 invented entities

The method relies on the standard density-clustering assumption, on a specific interpretation of sparse features as under-trained rather than noisy, and on the reliability of pseudo-labels. It also depends on several hand-set hyperparameters, some of which are not fully specified in the paper. There are no new physical entities or external benchmark predictions beyond the reported segmentation results.

free parameters (7)
  • Number of low-density anchors per class Nq = 256
    Set empirically in Implementation Details; controls how many sparse features are pulled toward cluster centers each batch.
  • Number of positive keys Np+ and negative keys Np- = 512 each
    Set empirically; balances the number of contrastive pairs and affects training stability.
  • Temperature parameter tau = 0.4
    Tuned via the sensitivity analysis in Fig. 5(d); affects the sharpness of the contrastive loss.
  • Mask threshold phi = not stated explicitly; sensitivity shown in Fig. 5(c)
    Used in Eq. 2 to binarize the mask; the default value is not given in Implementation Details.
  • Multi-scale neighbor graph scales {k1,...,kn} = not specified
    Eqs. 4-5 require a set of k values and a number of scales n, but the paper never states which values are used.
  • Memory bank size L = implied 1000 by Fig. 5(b)
    Bank size affects density estimation and positive key diversity; the implementation section does not state the default.
  • Loss weights lambda_cross and lambda_CL schedule = lambda_cross=1; lambda_CL=0.1*exp(-5*(1-t/tmax)^2)
    Chosen empirically; Fig. 5(a) shows sensitivity to these weights.
assumptions (3)
  • domain assumption Density-based clustering hypothesis: samples in the same class form high-density regions and decision boundaries lie in low-density areas.
    Invoked in the Introduction to justify extracting supervision from feature density; this is a standard SSL assumption but is not proven for medical image features.
  • ad hoc to paper Low-density features are under-trained or hard samples that should be pulled toward the cluster center, rather than outliers that should be ignored.
    This is the core assumption of the Soft Density-Guided Contrastive Learning module; if low density instead indicates noise, the loss would corrupt the feature space.
  • domain assumption Pseudo-labels and ground-truth masks are accurate enough to assign unlabeled features to the correct class for density estimation.
    Eq. 2 builds masks from pseudo-labels; in low-label regimes, pseudo-label errors propagate into density estimates and cluster centers.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neighbor Does Matter: Density-Aware Contrastive Learning for Medical Semi-supervised Segmentation." pith.science (2026). https://pith.science/paper/WNWUR52H

@misc{pith2026241219871,
  author       = {Pith},
  title        = {Pith review of: Neighbor Does Matter: Density-Aware Contrastive Learning for Medical Semi-supervised Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WNWUR52H}},
  note         = {Machine review of arXiv:2412.19871}
}
read the original abstract

In medical image analysis, multi-organ semi-supervised segmentation faces challenges such as insufficient labels and low contrast in soft tissues. To address these issues, existing studies typically employ semi-supervised segmentation techniques using pseudo-labeling and consistency regularization. However, these methods mainly rely on individual data samples for training, ignoring the rich neighborhood information present in the feature space. In this work, we argue that supervisory information can be directly extracted from the geometry of the feature space. Inspired by the density-based clustering hypothesis, we propose using feature density to locate sparse regions within feature clusters. Our goal is to increase intra-class compactness by addressing sparsity issues. To achieve this, we propose a Density-Aware Contrastive Learning (DACL) strategy, pushing anchored features in sparse regions towards cluster centers approximated by high-density positive samples, resulting in more compact clusters. Specifically, our method constructs density-aware neighbor graphs using labeled and unlabeled data samples to estimate feature density and locate sparse regions. We also combine label-guided co-training with density-guided geometric regularization to form complementary supervision for unlabeled data. Experiments on the Multi-Organ Segmentation Challenge dataset demonstrate that our proposed method outperforms state-of-the-art methods, highlighting its efficacy in medical image segmentation tasks.

Figures

Figures reproduced from arXiv: 2412.19871 by the authors.

Figure 1
Figure 1. Left (a): Qualitative comparison of abdominal organs. As observed in the red boxes, the contrastive learning method [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed unified learning framework. (a) shows the feature density-aware module. (b) shows our [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visualization of the segmentation results from different methods on the Synapse dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative results on Synapse dataset using dif [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Mean Dice performances on the Synapse dataset [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 53 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    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 capitalize " " * FUNCT...

  3. [3]

    Alonso, I.; Sabater, A.; Ferstl, D.; Montesano, L.; and Murillo, A. C. 2021. Semi-supervised semantic segmentation with pixel-level contrastive learning from a class-wise memory bank. In ICCV

  4. [4]

    Bai, Y.; Chen, D.; Li, Q.; Shen, W.; and Wang, Y. 2023. Bidirectional Copy-Paste for Semi-Supervised Medical Image Segmentation. In CVPR

  5. [5]

    Basak, H.; and Yin, Z. 2023. Pseudo-label guided contrastive learning for semi-supervised medical image segmentation. In CVPR

  6. [6]

    Bernard, O.; Lalande, A.; Zotti, C.; Cervenansky, F.; Yang, X.; Heng, P.-A.; Cetin, I.; Lekadir, K.; Camara, O.; Ballester, M. A. G.; et al. 2018. Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? TMI

  7. [7]

    Chaitanya, K.; Erdil, E.; Karani, N.; and Konukoglu, E. 2023. Local contrastive loss with pseudo-label based self-training for semi-supervised medical image segmentation. MIA

  8. [8]

    Chen, D.; Bai, Y.; Shen, W.; Li, Q.; Yu, L.; and Wang, Y. 2023 a . MagicNet: Semi-Supervised Multi-Organ Segmentation via Magic-Cube Partition and Recovery. In CVPR

Show all 64 references
  1. [9]

    Chen, T.; Kornblith, S.; Norouzi, M.; and Hinton, G. 2020. A simple framework for contrastive learning of visual representations. In ICML

  2. [10]

    Chen, X.; Yuan, Y.; Zeng, G.; and Wang, J. 2021. Semi-supervised semantic segmentation with cross pseudo supervision. In CVPR

  3. [11]

    Chen, Z.; Pan, Y.; and Xia, Y. 2023. Reconstruction-driven dynamic refinement based unsupervised domain adaptation for joint optic disc and cup segmentation. JBHI

  4. [12]

    Chen, Z.; Pan, Y.; Ye, Y.; Cui, H.; and Xia, Y. 2023 b . Treasure in distribution: a domain randomization based multi-source domain generalization for 2d medical image segmentation. In MICCAI

  5. [13]

    Chen, Z.; Pan, Y.; Ye, Y.; Lu, M.; and Xia, Y. 2024. Each test image deserves a specific prompt: Continual test-time adaptation for 2d medical image segmentation. In CVPR

  6. [14]

    Chi, H.; Pang, J.; Zhang, B.; and Liu, W. 2024. Adaptive Bidirectional Displacement for Semi-Supervised Medical Image Segmentation. In CVPR

  7. [15]

    L.; and Bouldin, D

    Davies, D. L.; and Bouldin, D. W. 1979. A cluster separation measure. TPAMI

  8. [16]

    P.-W.; Hu, X.; and Qin, J

    Deng, S.; Feng, Y.; Lin, H.; Fan, Y.; Lee, A. P.-W.; Hu, X.; and Qin, J. 2024. Semi-supervised TEE Segmentation via Interacting with SAM Equipped with Noise-Resilient Prompting. In AAAI

  9. [17]

    Dosovitskiy, A.; Beyer, L.; Kolesnikov, A.; Weissenborn, D.; Zhai, X.; Unterthiner, T.; Dehghani, M.; Minderer, M.; Heigold, G.; Gelly, S.; et al. 2020. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929

  10. [18]

    Gao, S.; Zhang, Z.; Ma, J.; Li, Z.; and Zhang, S. 2023. Correlation-aware mutual learning for semi-supervised medical image segmentation. In MICCAI

  11. [19]

    Hu, M.; Xia, P.; Wang, L.; Yan, S.; Tang, F.; Xu, Z.; Luo, Y.; Song, K.; Leitner, J.; Cheng, X.; et al. 2025. Ophnet: A large-scale video benchmark for ophthalmic surgical workflow understanding. In ECCV

  12. [20]

    Hu, M.; Yuan, K.; Shen, Y.; Tang, F.; Xu, X.; Zhou, L.; Li, W.; Chen, Y.; Xu, Z.; Peng, Z.; et al. 2024. OphCLIP: Hierarchical Retrieval-Augmented Learning for Ophthalmic Surgical Video-Language Pretraining. arXiv preprint arXiv:2411.15421

  13. [21]

    Huang, H.; Huang, Y.; Xie, S.; Lin, L.; Tong, R.; Chen, Y.-W.; Li, Y.; and Zheng, Y. 2024. Combinatorial CNN-Transformer Learning with Manifold Constraints for Semi-supervised Medical Image Segmentation. In AAAI

  14. [22]

    Jiang, S.; Wu, H.; Chen, J.; Zhang, Q.; and Qin, J. 2024. PH-Net: Semi-Supervised Breast Lesion Segmentation via Patch-wise Hardness. In CVPR

  15. [23]

    Khosla, P.; Teterwak, P.; Wang, C.; Sarna, A.; Tian, Y.; Isola, P.; Maschinot, A.; Liu, C.; and Krishnan, D. 2020. Supervised contrastive learning. NeurIPS

  16. [24]

    E.; Styner, M.; Langerak, T

    Landman, B.; Xu, Z.; Igelsias, J. E.; Styner, M.; Langerak, T. R.; and Klein, A. 2015. 2015 miccai multi-atlas labeling beyond the cranial vault workshop and challenge. In MICCAI Multi-Atlas Labeling Beyond Cranial Vault—Workshop Challenge

  17. [25]

    Li, J.; Yu, H.; Chen, C.; Ding, M.; and Zha, S. 2022. Category guided attention network for brain tumor segmentation in MRI. Physics in Medicine & Biology, 67(8): 085014

  18. [26]

    Liu, X.; Li, W.; and Yuan, Y. 2024. DiffRect: Latent Diffusion Label Rectification for Semi-supervised Medical Image Segmentation. In MICCAI

  19. [27]

    Liu, Y.; Zhu, H.; Liu, M.; Yu, H.; Chen, Z.; and Gao, J. 2024. Rolling-Unet: Revitalizing MLP’s Ability to Efficiently Extract Long-Distance Dependencies for Medical Image Segmentation. In AAAI

  20. [28]

    Liu, Z.; Ma, Q.; Ma, P.; and Wang, L. 2023. Temporal-frequency co-training for time series semi-supervised learning. In AAAI

  21. [29]

    Liu, Z.; Wu, S.; Xu, C.; Wang, X.; Zhu, L.; Wu, S.; and Feng, F. 2022. Copy motion from one to another: Fake motion video generation. arXiv preprint arXiv:2205.01373

  22. [30]

    Luo, X.; Liao, W.; Chen, J.; Song, T.; Chen, Y.; Zhang, S.; Chen, N.; Wang, G.; and Zhang, S. 2021. Efficient semi-supervised gross target volume of nasopharyngeal carcinoma segmentation via uncertainty rectified pyramid consistency. In MICCAI

  23. [31]

    Ma, Q.; Zhang, J.; Qi, L.; Yu, Q.; Shi, Y.; and Gao, Y. 2024. Constructing and Exploring Intermediate Domains in Mixed Domain Semi-supervised Medical Image Segmentation. In CVPR

  24. [32]

    Miao, J.; Zhou, S.-P.; Zhou, G.-Q.; Wang, K.-N.; Yang, M.; Zhou, S.; and Chen, Y. 2023. SC-SSL: Self-correcting Collaborative and Contrastive Co-training Model for Semi-Supervised Medical Image Segmentation. TMI

  25. [33]

    Mofid, B.; Mosalla, S. M. M.; Goodarzi, M.; and Tavakoli, H. 2024. Deep CNN-based Fully Automated Segmentation of Pelvic Multi-Organ on CT Images for Prostate Cancer Radiotherapy. Journal of Biomedical Physics and Engineering

  26. [34]

    Nguyen-Duc, T.; Le, T.; Bammer, R.; Zhao, H.; Cai, J.; and Phung, D. 2023. Cross-adversarial local distribution regularization for semi-supervised medical image segmentation. In MICCAI

  27. [35]

    L.; Zhou, Z.; et al

    Qu, C.; Zhang, T.; Qiao, H.; Tang, Y.; Yuille, A. L.; Zhou, Z.; et al. 2024. Abdomenatlas-8k: Annotating 8,000 CT volumes for multi-organ segmentation in three weeks. NeurIPS

  28. [36]

    Rosenberg, A.; and Hirschberg, J. 2007. V-measure: A conditional entropy-based external cluster evaluation measure. In EMNLP-CoNLL

  29. [37]

    Rousseeuw, P. J. 1987. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of computational and applied mathematics

  30. [38]

    N.; and Jagersand, M

    Siam, M.; Oreshkin, B. N.; and Jagersand, M. 2019. Amp: Adaptive masked proxies for few-shot segmentation. In ICCV

  31. [39]

    Tang, F.; Trinh, M.; Duong, A.; Ly, A.; Stapleton, F.; Chen, Z.; Ge, Z.; and Razzak, I. 2024 a . Discriminating retinal microvascular and neuronal differences related to migraines: Deep Learning based Crossectional Study. arXiv preprint arXiv:2408.07293

  32. [40]

    Tang, F.; Xu, Z.; Huang, Q.; Wang, J.; Hou, X.; Su, J.; and Liu, J. 2023. DuAT: Dual-aggregation transformer network for medical image segmentation. In PRCV

  33. [41]

    Tang, F.; Xu, Z.; Qu, Z.; Feng, W.; Jiang, X.; and Ge, Z. 2024 b . Hunting Attributes: Context Prototype-Aware Learning for Weakly Supervised Semantic Segmentation. In CVPR

  34. [42]

    Van der Maaten, L.; and Hinton, G. 2008. Visualizing data using t-SNE. Journal of machine learning research

  35. [43]

    Wang, H.; and Li, X. 2023. Dhc: Dual-debiased heterogeneous co-training framework for class-imbalanced semi-supervised medical image segmentation. In MICCAI

  36. [44]

    Wang, J.; Huang, Q.; Tang, F.; Meng, J.; Su, J.; and Song, S. 2022. Stepwise feature fusion: Local guides global. In MICCAI. Springer

  37. [45]

    Wang, X.; Zhang, B.; Yu, L.; and Xiao, J. 2023. Hunting Sparsity: Density-Guided Contrastive Learning for Semi-Supervised Semantic Segmentation. In CVPR

  38. [46]

    Wang, Y.; Zhou, W.; Jiang, T.; Bai, X.; and Xu, Y. 2020. Intra-class feature variation distillation for semantic segmentation. In ECCV

  39. [47]

    E.; Rakic, M.; Guttag, J.; and Dalca, A

    Wong, H. E.; Rakic, M.; Guttag, J.; and Dalca, A. V. 2024. Scribbleprompt: Fast and flexible interactive segmentation for any medical image. In ECCV

  40. [48]

    Wu, L.; Zhuang, J.; and Chen, H. 2024. Voco: A simple-yet-effective volume contrastive learning framework for 3d medical image analysis. In CVPR

  41. [49]

    Wu, S.; Chen, H.; Yin, Y.; Hu, S.; Feng, R.; Jiao, Y.; Yang, Z.; and Liu, Z. 2024 a . Joint-Motion Mutual Learning for Pose Estimation in Video. In ACMMM

  42. [50]

    Wu, S.; Liu, Z.; Zhang, B.; Zimmermann, R.; Ba, Z.; Zhang, X.; and Ren, K. 2024 b . Do as I Do: Pose Guided Human Motion Copy. IEEE Transactions on Dependable and Secure Computing

  43. [51]

    Wu, Y.; Ge, Z.; Zhang, D.; Xu, M.; Zhang, L.; Xia, Y.; and Cai, J. 2022 a . Mutual consistency learning for semi-supervised medical image segmentation. MIA

  44. [52]

    Wu, Y.; Wu, Z.; Wu, Q.; Ge, Z.; and Cai, J. 2022 b . Exploring smoothness and class-separation for semi-supervised medical image segmentation. In MICCAI

  45. [53]

    Wu, Y.; Xu, M.; Ge, Z.; Cai, J.; and Zhang, L. 2021. Semi-supervised left atrium segmentation with mutual consistency training. In MICCAI

  46. [54]

    Xu, C.; Zhang, T.; Zhang, D.; Zhang, D.; and Han, J. 2024 a . Deep Generative Adversarial Reinforcement Learning for Semi-Supervised Segmentation of Low-Contrast and Small Objects in Medical Images. TMI

  47. [55]

    Xu, Z.; Tang, F.; Chen, Z.; Zhou, Z.; Wu, W.; Yang, Y.; Liang, Y.; Jiang, J.; Cai, X.; and Su, J. 2024 b . Polyp-Mamba: Polyp Segmentation with Visual Mamba. In MICCAI

  48. [56]

    Yang, Y.; Chen, H.; Liu, Z.; Lyu, Y.; Zhang, B.; Wu, S.; Wang, Z.; and Ren, K. 2023. Action recognition with multi-stream motion modeling and mutual information maximization. arXiv preprint arXiv:2306.07576

  49. [57]

    Ye, Y.; Xie, Y.; Zhang, J.; Chen, Z.; Wu, Q.; and Xia, Y. 2024 a . Continual self-supervised learning: Towards universal multi-modal medical data representation learning. In CVPR

  50. [58]

    Ye, Y.; Zhang, J.; Chen, Z.; and Xia, Y. 2022. Desd: Self-supervised learning with deep self-distillation for 3d medical image segmentation. In MICCAI

  51. [59]

    Ye, Y.; Zhang, J.; Chen, Z.; and Xia, Y. 2024 b . CADS: A Self-supervised Learner via Cross-modal Alignment and Deep Self-distillation for CT Volume Segmentation. TMI

  52. [60]

    You, C.; Dai, W.; Min, Y.; Staib, L.; Sekhon, J.; and Duncan, J. S. 2023. Action++: Improving semi-supervised medical image segmentation with adaptive anatomical contrast. In MICCAI

  53. [61]

    Yu, L.; Wang, S.; Li, X.; Fu, C.-W.; and Heng, P.-A. 2019. Uncertainty-aware self-ensembling model for semi-supervised 3D left atrium segmentation. In MICCAI

  54. [62]

    Zhao, X.; Tang, F.; Wang, X.; and Xiao, J. 2024 a . Sfc: Shared feature calibration in weakly supervised semantic segmentation. In AAAI

  55. [63]

    Zhao, Z.; Wang, Z.; Wang, L.; Yuan, Y.; and Zhou, L. 2024 b . Alternate Diverse Teaching for Semi-supervised Medical Image Segmentation. In ECCV

  56. [64]

    Zheng, Z.; Hayashi, Y.; Oda, M.; Kitasaka, T.; Misawa, K.; and Mori, K. 2024. Federated 3D multi-organ segmentation with partially labeled and unlabeled data. International Journal of Computer Assisted Radiology and Surgery

Pith tools

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