Pith. sign in

REVIEW 3 major objections 4 minor 46 references

SuperCL: Superpixel Guided Contrastive Learning for Medical Image Segmentation Pre-training

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read SuperCL pre-trains a U-Net with SLIC superpixel pseudo masks to define positive contrastive pairs, and it outperforms 12 methods on 8 medical segmentation datasets under limited annotations.

desk verdict Superpixel-guided contrastive pre-training that looks genuinely useful, but the reported SOTA gains rest on test-set hyperparameter tuning and unstated seed variance; worth a careful peer review, not yet a claim to trust. read the letter →

arxiv 2504.14737 v1 pith:VP5RWTCU submitted 2025-04-20 cs.CV cs.AI

classification cs.CVcs.AI
keywords contrastivelearningmedicalimagesegmentationsuperpixelself-supervisedpre-traininglimitedannotationsSLICU-Netintra-imagepairs
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 proposes SuperCL, a contrastive pre-training method for medical image segmentation when annotated data are scarce. Its central idea is to let superpixel maps act as pseudo labels: pixels that fall in the same SLIC superpixel are treated as positive pairs for an intra-image contrastive loss, and averaged superpixel features feeding a connected-components step on a nearest-neighbor graph define positive pairs across images. This design removes the manually set thresholds that earlier medical contrastive methods rely on. The paper reports that SuperCL outperforms 12 baselines on 8 CT and MRI datasets, with DSC gains of 3.15, 5.44, and 7.89 points over the previous best on MMWHS, CHAOS, and Spleen at 10% annotations.

What carries the argument

The load-bearing object is the superpixel pseudo mask $S \in \mathbb{R}^{B \times h \times w}$, generated by SLIC and downsampled to feature-map resolution. Pixel-wise, it defines the positive set $\Omega_{ILCP}$ for the intra-image loss $L_{intra}$ in Eq. (5). Instance-wise, its per-cluster averaged features $P_c$ feed the ASP module, and the connected-components weak label $W$ from a top-1 nearest-neighbor graph defines the positive set for $L_{inter}$ in Eq. (9). The machinery converts raw pixel intensities into a grouping prior that substitutes for manually tuned thresholds.

What would settle it

Corrupt the superpixel pseudo mask in a controlled way, for example by randomly reassigning 10–50% of superpixel labels or by shifting boundaries away from organ edges, and re-run pre-training with identical settings; if downstream DSC on MMWHS and CHAOS does not drop as corruption increases, the reported gains are not caused by the superpixel grouping itself.

Watch

Extended reading notes

Core claim

The central claim is that superpixel structure is a reliable, annotation-free grouping signal for contrastive learning in medical images. SuperCL computes a superpixel map per image, flattens it into a pseudo mask aligned with the encoder feature map, and applies a supervised contrastive loss that pulls together features of pixels sharing a superpixel cluster (ILCP, Eq. (5)). For inter-image pairs, it averages the feature map inside each superpixel (ASP) and runs connected-components labeling on a top-1 nearest-neighbor graph over those averaged features (CCL), producing weak labels that define global positives (IGCP, Eq. (9)). The paper argues this exploits both intra-image pixel correlation and inter-image structural similarity while avoiding threshold tuning, and that it yields state-of-the-art segmentation performance when a U-Net is fine-tuned on 10% or 25% of downstream labels.

Load-bearing premise

The load-bearing premise is that SLIC superpixels, computed from raw pixel intensities, group pixels into the same anatomical region, so that pixels sharing a superpixel are genuinely positive pairs rather than false ones.

Editorial extensions

If this is right

  • If the reported gains hold, pre-training on unlabeled CT/MRI volumes with SuperCL should improve U-Net segmentation at 10% and 25% annotation rates across multi-organ and ROI tasks, with the largest margins at the lowest label ratios.
  • The same pre-trained encoder transfers to U-Net variants (AttUNet, UCTransNet, BCDUNet, ResUNet, RollingUNet, UKAN), so the benefit is not tied to one decoder design.
  • SuperCL also serves as an initialization for semi-supervised pipelines: the paper reports that it raises BCP and CauSSL performance on ACDC at 10% and 20% labels.
  • Because the pseudo masks come from a generic algorithm, the strategy can be dropped into other dense-prediction encoders as long as a feature map can be aligned with the superpixel map.

Reading between the lines

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

  • The method's ceiling is set by superpixel boundary quality: on lesions or organs with low intensity contrast, SLIC clusters will cross anatomical borders, and ILCP will then pull together features that should stay apart; comparing SLIC with boundary-aware superpixels would show whether segmentation gains track edge recall.
  • SuperCL is threshold-free only for the pairing step; cluster number, stride, and loss weights still need tuning, and the paper reports sensitivity only for cluster number and stride, not for all eight datasets.
  • ASP performs a fixed non-learnable average pooling inside each superpixel; making that pooling learnable or multi-scale is a natural extension that could strengthen the inter-image positives.
  • The semi-supervised experiments cover only ACDC; testing SuperCL as an initialization for semi-supervised methods on CT datasets would tell whether the benefit generalizes across modalities.
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 / 4 minor

Summary. The paper proposes SuperCL, a self-supervised contrastive pre-training method for medical image segmentation. SuperCL uses SLIC superpixels to define two contrastive pair generation strategies: ILCP, which treats pixels within the same superpixel as positive pairs in an intra-image pixel-level contrastive loss, and IGCP, which uses ASP (averaging features per superpixel region) and CCL (connected components on a 1-nearest-neighbor graph) to build a weak label for an inter-image instance-level contrastive loss. The method is evaluated by pre-training a U-Net encoder on CHD, BraTS2018, and KiTS2019, then fine-tuning on eight downstream datasets (ACDC, MMWHS, HVSMR, CHAOS, Spleen, ISIC, Heart, Hippocampus) with 10% and 25% of the training labels, comparing against 12 baselines. The authors report consistent DSC improvements and additionally show that SuperCL can initialize semi-supervised methods (BCP, CauSSL), with ablations isolating the contributions of ILCP and IGCP and sensitivity studies on stride and superpixel cluster number.

Significance. If the reported gains are robust, SuperCL would be a useful addition to the medical-image contrastive-learning toolbox: it replaces manually tuned thresholds for contrastive pair generation with a superpixel-derived grouping, and it is evaluated far more broadly than is typical in this area (8 downstream datasets, 12 baselines, ablations, backbone generalization, and semi-supervised integration). The proposed modules (ASP and CCL) are simple, plausible, and clearly described. The main unresolved issue is not the method's internal logic but the strength of the empirical evidence supporting the central 'outperforms 12 methods' claim, given the evaluation protocol described below.

major comments (3)
  1. [Table IV and Fig. 8] The same concern applies to the loss weights (lambda_1, lambda_2, lambda_3 = 1.0, 1.0, 0.5) and temperature (tau = 0.1) in Section III-A and the Implementation Details: no sensitivity analysis or validation-based selection is described for these values, and if they were chosen with knowledge of the test-set results, the reported gains are again subject to selection bias.
  2. [Section IV-D (Implementation Details)] This issue is load-bearing for the abstract's central claim, which is stated as a definitive comparison ('outperforms existing 12 methods'). At minimum, the manuscript should present error bars for the key comparisons in Tables I and II, or clearly state that the reported values are single runs and therefore only indicative.
  3. [Section III-B, Eq. (5)] The claim that superpixels provide semantically valid positive pairs is the load-bearing assumption of ILCP, but the paper does not provide any direct evidence that the SLIC superpixel boundaries align with anatomical or tissue boundaries in the pre-training datasets, nor does it quantify the sensitivity of the result to this assumption. Fig. 6 and Fig. 8 show the effect of superpixel method and cluster count on downstream DSC, which is indirect evidence, but it would strengthen the paper to report, for a few representative slices, the fraction of superpixels that straddle label boundaries (using the available pre-training annotations) or otherwise to test the robustness of the method to the superpixel granularity. The current ablation varies K from 1 to 1024, which is useful, but the interpretation of the results as evidence for boundary alignment is not established.
minor comments (4)
  1. [Throughout] The manuscript contains numerous typos and grammatical errors that should be corrected before publication, including 'avaliable', 'Resonably', 'performace', 'simlar', 'avereged', 'untilizing', 'pesudo', 'datset', and 'Simens' (e.g., in Sections I, III-C, IV-A, IV-B, and the dataset descriptions).
  2. [Section III-B] The text refers to 'Section 2.2' and 'Section 2.3' when discussing ILCP and IGCP, but the actual sections are III-B and III-C; please update the cross-references.
  3. [Section III-C, Eq. (6)] The definition of the averaged superpixel feature P is ambiguous: each P_c is a scalar (the mean of the channel-averaged feature over superpixel region c), yet later P is used as an image of size h x w. Please clarify that P is formed by filling each superpixel region with the scalar value P_c, and state explicitly which image (the original X or the augmented view I_1/I_2) is used to compute the SLIC superpixel map S, since non-spatial augmentations change pixel intensities and could affect the superpixel grouping.
  4. [Section IV-B, Fig. 8] The statement in the discussion of Fig. 8 that 'our SuperCL among all the cluster numbers except for 150 outperforms the PCL baseline' would be more informative if the figure showed error bars or repeated runs, since the differences among many cluster numbers appear small (within about 1% DSC); as written, the figure may not support the conclusion that K = 100 is uniquely or significantly better.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the superpixel pseudo-labels and weak labels are generated from inputs and internal features, and downstream metrics are held out; the lone LNSNet self-citation is not load-bearing.

full rationale

SuperCL's two contrastive objectives are defined directly from inputs and internal representations, not from the downstream labels or the reported metrics. ILCP (Eq. 5) uses superpixel masks produced by SLIC on each input image to define positive pixel pairs; L_intra is the standard supervised InfoNCE loss with those pseudo-labels, so the pre-training target is a function of the input image and the current encoder, not of segmentation accuracy. IGCP (Eqs. 6-9) builds weak labels by averaging features inside superpixels (ASP) and taking connected components of a nearest-neighbor graph (CCL); again, the pseudo-label is generated from the model's own features and the SLIC map. The downstream DSC/JC/HD95/ASD numbers are measured on a fixed 8:2 test split after fine-tuning, and the same protocol is used for all baselines; no reported number is defined as a fit to those numbers. The only self-citation, LNSNet [27] (co-authored by Lei Zhu and Yanye Lu), appears in the related-work survey and in a superpixel-method comparison (Fig. 6), where SLIC is the actual generator used by SuperCL; that citation is not load-bearing. Section IV-B reports test-set hyperparameter selection (K=100 and stride=1 chosen on ACDC/MMWHS with 25% labels), which is a correctness and overfitting concern rather than a circular reduction: the reported metrics are not equal by construction to the selection criterion, and the same settings are applied to other datasets and label ratios. No circular step satisfying the quoted-equation or fitted-parameter-renamed-as-prediction standard was found, so the derivation chain is self-contained.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities; the method introduces loss formulations and modules (ASP, CCL) rather than new ontological objects.

free parameters (4)
  • loss weights lambda_1, lambda_2, lambda_3 = 1.0, 1.0, 0.5
    Empirically set in pre-training; not ablated for sensitivity.
  • temperature tau = 0.1
    Standard contrastive temperature chosen empirically.
  • number of superpixel clusters K = 100
    Selected as best from gradient experiment in Fig. 8 on ACDC and MMWHS with 25% labels; used in all main results.
  • stride for ILCP pixel sampling = 1
    Selected as best from Table IV sensitivity study; all comparisons and ablations use stride 1.
assumptions (4)
  • domain assumption Superpixel clusters correspond to semantically coherent regions
    Section III-B: pixels from same superpixel cluster are treated as positive pairs; if clusters cross tissue boundaries, the loss enforces incorrect grouping.
  • domain assumption Connected components of the top-1 nearest neighbor graph define meaningful positive image groups
    Section III-C, Eq. 7-8, adopted from WCL; if the graph is dominated by trivial intensity similarities, weak labels are noisy.
  • domain assumption Pre-training on upstream unlabeled datasets transfers to downstream modalities
    Section IV-A: KiTS (CT) pre-training is used for ISIC (dermoscopy) fine-tuning; this transfer is assumed beneficial without separate validation.
  • standard math InfoNCE / supervised contrastive loss is a valid objective
    Standard background.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SuperCL: Superpixel Guided Contrastive Learning for Medical Image Segmentation Pre-training." pith.science (2026). https://pith.science/paper/VP5RWTCU

@misc{pith2026250414737,
  author       = {Pith},
  title        = {Pith review of: SuperCL: Superpixel Guided Contrastive Learning for Medical Image Segmentation Pre-training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VP5RWTCU}},
  note         = {Machine review of arXiv:2504.14737}
}
read the original abstract

Medical image segmentation is a critical yet challenging task, primarily due to the difficulty of obtaining extensive datasets of high-quality, expert-annotated images. Contrastive learning presents a potential but still problematic solution to this issue. Because most existing methods focus on extracting instance-level or pixel-to-pixel representation, which ignores the characteristics between intra-image similar pixel groups. Moreover, when considering contrastive pairs generation, most SOTA methods mainly rely on manually setting thresholds, which requires a large number of gradient experiments and lacks efficiency and generalization. To address these issues, we propose a novel contrastive learning approach named SuperCL for medical image segmentation pre-training. Specifically, our SuperCL exploits the structural prior and pixel correlation of images by introducing two novel contrastive pairs generation strategies: Intra-image Local Contrastive Pairs (ILCP) Generation and Inter-image Global Contrastive Pairs (IGCP) Generation. Considering superpixel cluster aligns well with the concept of contrastive pairs generation, we utilize the superpixel map to generate pseudo masks for both ILCP and IGCP to guide supervised contrastive learning. Moreover, we also propose two modules named Average SuperPixel Feature Map Generation (ASP) and Connected Components Label Generation (CCL) to better exploit the prior structural information for IGCP. Finally, experiments on 8 medical image datasets indicate our SuperCL outperforms existing 12 methods. i.e. Our SuperCL achieves a superior performance with more precise predictions from visualization figures and 3.15%, 5.44%, 7.89% DSC higher than the previous best results on MMWHS, CHAOS, Spleen with 10% annotations. Our code will be released after acceptance.

Figures

Figures reproduced from arXiv: 2504.14737 by the authors.

Figure 1
Figure 1. Our SuperCL (solid red lines) achieves SOTA segmentation perfor [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of our proposed SuperCL. The image is input into two branches with two different augmentation settings: (1) The spatial invariance group [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Superpixel-guided intra-image local contrastive pairs generation. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Illustration of our proposed ASP and CCL modules. (a) ASP aims at generating a more reliable representation for affinity matrix calculation. (b) CCL [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Visualization of multi-organ segmentation results on ACDC, MMWHS, CHAOS and HVSMR. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Comparison results of different superpixel methods on ACDC and [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Comparison results of different variants of UNet backbone on ACDC [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Comparison results of different superpixel cluster numbers on ACDC and MMWHS with 25% annotations. [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: The performance (DSC) of each individual organ on ACDC and MMWHS with 10% annotations. [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: The performance (DSC) of each individual organ on HVSMR and CHAOS with 10% annotations. [PITH_FULL_IMAGE:figures/full_fig_p014_10.png]
Figure 11
Figure 11. Figure 11: Visualization of more multi-organ segmentation results on ACDC, MMWHS, CHAOS and HVSMR. [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 12
Figure 12. Figure 12: Visualization of ROI-based segmentation results on Spleen, Heart and Spleen. [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 32 canonical work pages

  1. [1]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in International conference on machine learning . PMLR, 2020, pp. 1597–1607. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 11

  2. [2]

    Momentum contrast for unsupervised visual representation learning,

    K. He, H. Fan, Y . Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in Proceedings of the CVF conference on computer vision and pattern recognition , 2020, pp. 9729–9738

  3. [3]

    Contrastive learning of global and local features for medical image segmentation with limited annotations,

    K. Chaitanya, E. Erdil, N. Karani, and E. Konukoglu, “Contrastive learning of global and local features for medical image segmentation with limited annotations,” Advances in Neural Information Processing Systems, vol. 33, pp. 12 546–12 558, 2020

  4. [4]

    Positional contrastive learning for volumetric medical image segmentation,

    D. Zeng, Y . Wu, X. Hu, X. Xu, H. Yuan, M. Huang, J. Zhuang, J. Hu, and Y . Shi, “Positional contrastive learning for volumetric medical image segmentation,” in International Conference on Medical Image Computing and Computer-Assisted Intervention . Springer, 2021, pp. 221–230

  5. [5]

    Dira: Discriminative, restorative, and adversarial learning for self-supervised medical image analysis,

    F. Haghighi, M. R. H. Taher, M. B. Gotway, and J. Liang, “Dira: Discriminative, restorative, and adversarial learning for self-supervised medical image analysis,” in Proceedings of the CVF Conference on Computer Vision and Pattern Recognition , June 2022, pp. 20 824– 20 834

  6. [6]

    Unsupervised learning of visual features by contrasting cluster assign- ments,

    M. Caron, I. Misra, J. Mairal, P. Goyal, P. Bojanowski, and A. Joulin, “Unsupervised learning of visual features by contrasting cluster assign- ments,” Advances in neural information processing systems , vol. 33, pp. 9912–9924, 2020

  7. [7]

    Bootstrap your own latent-a new approach to self-supervised learning,

    J.-B. Grill, F. Strub, F. Altch ´e, C. Tallec, P. Richemond, E. Buchatskaya, C. Doersch, B. Avila Pires, Z. Guo, M. Gheshlaghi Azar et al. , “Bootstrap your own latent-a new approach to self-supervised learning,” Advances in neural information processing systems, vol. 33, pp. 21 271– 21 284, 2020

  8. [8]

    Exploring simple siamese representation learning,

    X. Chen and K. He, “Exploring simple siamese representation learning,” in Proceedings of the CVF conference on computer vision and pattern recognition, 2021, pp. 15 750–15 758

Show all 46 references
  1. [9]

    Weakly supervised contrastive learning,

    M. Zheng, F. Wang, S. You, C. Qian, C. Zhang, X. Wang, and C. Xu, “Weakly supervised contrastive learning,” in Proceedings of the CVF International Conference on Computer Vision, 2021, pp. 10 042–10 051

  2. [10]

    Desd: Self-supervised learning with deep self-distillation for 3d medical image segmentation,

    Y . Ye, J. Zhang, Z. Chen, and Y . Xia, “Desd: Self-supervised learning with deep self-distillation for 3d medical image segmentation,” in Med- ical Image Computing and Computer Assisted Intervention – MICCAI 2022, L. Wang, Q. Dou, P. T. Fletcher, S. Speidel, and S. Li, Eds. C...

  3. [11]

    Simtriplet: Simple triplet representation learning with a single gpu,

    Q. Liu, P. C. Louis, Y . Lu, A. Jha, M. Zhao, R. Deng, T. Yao, J. T. Roland, H. Yang, S. Zhaoet al., “Simtriplet: Simple triplet representation learning with a single gpu,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2021: 24th International Conference...

  4. [12]

    Slic superpixels compared to state-of-the-art superpixel methods,

    R. Achanta, A. Shaji, K. Smith, A. Lucchi, P. Fua, and S. S ¨usstrunk, “Slic superpixels compared to state-of-the-art superpixel methods,”IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 34, no. 11, pp. 2274–2282, 2012

  5. [13]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international con- ference, Munich, Germany, October 5-9, 2015, proceedings, part III 18 ...

  6. [14]

    Atten- tion u-net: Learning where to look for the pancreas,

    O. Oktay, J. Schlemper, L. L. Folgoc, M. Lee, M. Heinrich, K. Misawa, K. Mori, S. McDonagh, N. Y . Hammerla, B. Kainz et al. , “Atten- tion u-net: Learning where to look for the pancreas,” arXiv preprint arXiv:1804.03999, 2018

  7. [15]

    Uctransnet: rethinking the skip connections in u-net from a channel-wise perspective with trans- former,

    H. Wang, P. Cao, J. Wang, and O. R. Zaiane, “Uctransnet: rethinking the skip connections in u-net from a channel-wise perspective with trans- former,” in Proceedings of the AAAI conference on artificial intelligence, vol. 36, no. 3, 2022, pp. 2441–2449

  8. [16]

    Bi- directional convlstm u-net with densley connected convolutions,

    R. Azad, M. Asadi-Aghbolaghi, M. Fathy, and S. Escalera, “Bi- directional convlstm u-net with densley connected convolutions,” in Proceedings of the CVF international conference on computer vision workshops, 2019, pp. 0–0

  9. [17]

    The importance of skip connections in biomedical image segmentation,

    M. Drozdzal, E. V orontsov, G. Chartrand, S. Kadoury, and C. Pal, “The importance of skip connections in biomedical image segmentation,” in International workshop on deep learning in medical image analysis . Springer, 2016, pp. 179–187

  10. [18]

    Rolling-unet: Re- vitalizing mlp’s ability to efficiently extract long-distance dependencies for medical image segmentation,

    Y . Liu, H. Zhu, M. Liu, H. Yu, Z. Chen, and J. Gao, “Rolling-unet: Re- vitalizing mlp’s ability to efficiently extract long-distance dependencies for medical image segmentation,” Proceedings of the AAAI Conference on Artificial Intelligence , vol. 38, no. 4, pp. 3819–3827, Mar. 2024

  11. [19]

    U-kan makes strong backbone for medical image segmentation and generation,

    C. Li, X. Liu, W. Li, C. Wang, H. Liu, Y . Liu, Z. Chen, and Y . Yuan, “U-kan makes strong backbone for medical image segmentation and generation,” 2024

  12. [20]

    Superpixels and polygons using simple non-iterative clustering,

    R. Achanta and S. Susstrunk, “Superpixels and polygons using simple non-iterative clustering,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , July 2017

  13. [21]

    Superpixel segmentation using linear spectral clustering,

    Z. Li and J. Chen, “Superpixel segmentation using linear spectral clustering,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 2015

  14. [22]

    Efficient Graph-Based image segmentation,

    P. F. Felzenszwalb and D. P. Huttenlocher, “Efficient Graph-Based image segmentation,” Int. J. Comput. Vis. , vol. 59, no. 2, pp. 167–181, Sep. 2004

  15. [23]

    Entropy rate superpixel segmentation,

    M.-Y . Liu, O. Tuzel, S. Ramalingam, and R. Chellappa, “Entropy rate superpixel segmentation,” in CVPR 2011, 2011, pp. 2097–2104

  16. [24]

    Learning superpixels with segmentation-aware affinity loss,

    W.-C. Tu, M.-Y . Liu, V . Jampani, D. Sun, S.-Y . Chien, M.-H. Yang, and J. Kautz, “Learning superpixels with segmentation-aware affinity loss,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 2018

  17. [25]

    Superpixel sampling networks,

    V . Jampani, D. Sun, M.-Y . Liu, M.-H. Yang, and J. Kautz, “Superpixel sampling networks,” in Proceedings of the European Conference on Computer Vision, September 2018

  18. [26]

    Superpixel segmentation with fully convolutional networks,

    F. Yang, Q. Sun, H. Jin, and Z. Zhou, “Superpixel segmentation with fully convolutional networks,” in Proceedings of the CVF Conference on Computer Vision and Pattern Recognition , June 2020

  19. [27]

    Learning the superpixel in a non-iterative and lifelong manner,

    L. Zhu, Q. She, B. Zhang, Y . Lu, Z. Lu, D. Li, and J. Hu, “Learning the superpixel in a non-iterative and lifelong manner,” in Proceedings of the CVF Conference on Computer Vision and Pattern Recognition , June 2021, pp. 1225–1234

  20. [28]

    Whole heart and great vessel segmentation in congenital heart disease using deep neural networks and graph matching,

    X. Xu, T. Wang, Y . Shi, H. Yuan, Q. Jia, M. Huang, and J. Zhuang, “Whole heart and great vessel segmentation in congenital heart disease using deep neural networks and graph matching,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2019: 22nd Internation...

  21. [29]

    Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challenge,

    S. Bakas, M. Reyes, A. Jakab, S. Bauer, M. Rempfler, A. Crimi, R. T. Shinohara, C. Berger, S. M. Ha, M. Rozycki et al., “Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challeng...

  22. [30]

    The state of the art in kidney and kidney tumor segmentation in contrast-enhanced ct imaging: Results of the kits19 challenge,

    N. Heller, F. Isensee, K. H. Maier-Hein, X. Hou, C. Xie, F. Li, Y . Nan, G. Mu, Z. Lin, M. Han et al. , “The state of the art in kidney and kidney tumor segmentation in contrast-enhanced ct imaging: Results of the kits19 challenge,” Medical image analysis, vol. 67, p. 101821, 2021

  23. [31]

    Deep learning techniques for automatic mri cardiac multi-structures segmentation and diagnosis: Is the problem solved?

    O. Bernard, A. Lalande, Zotti, and et al., “Deep learning techniques for automatic mri cardiac multi-structures segmentation and diagnosis: Is the problem solved?” IEEE Transactions on Medical Imaging , vol. 37, no. 11, pp. 2514–2525, 2018

  24. [32]

    Challenges and methodologies of fully automatic whole heart segmentation: a review,

    X. Zhuang, “Challenges and methodologies of fully automatic whole heart segmentation: a review,” Journal of healthcare engineering, vol. 4, no. 3, pp. 371–407, 2013

  25. [33]

    Interactive whole-heart segmentation in congenital heart disease,

    D. F. Pace, A. V . Dalca, T. Geva, A. J. Powell, M. H. Moghari, and P. Golland, “Interactive whole-heart segmentation in congenital heart disease,” in Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Proceed- ings, Part III...

  26. [34]

    CHAOS Challenge - combined (CT-MR) healthy abdominal organ segmentation,

    A. E. Kavur, N. S. Gezer, M. Barıs ¸, S. Aslan, P.-H. Conze, V . Groza, D. D. Pham, S. Chatterjee, P. Ernst, S. ¨Ozkan, B. Baydar, D. Lachinov, S. Han, J. Pauli, F. Isensee, M. Perkonigg, R. Sathish, R. Rajan, D. Sheet, G. Dovletov, O. Speck, A. N ¨urnberger, K. H. Maier-Hein,...

  27. [35]

    The medical segmentation decathlon,

    M. Antonelli, A. Reinke, S. Bakas, K. Farahani, A. Kopp-Schneider, B. A. Landman, G. Litjens, B. Menze, O. Ronneberger, R. M. Summers et al. , “The medical segmentation decathlon,” Nature communications, vol. 13, no. 1, p. 4128, 2022

  28. [36]

    Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (ISIC),

    N. C. F. Codella, V . Rotemberg, P. Tschandl, M. E. Celebi, S. W. Dusza, D. A. Gutman, B. Helba, A. Kalloo, K. Liopyris, M. A. Marchetti, H. Kittler, and A. Halpern, “Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging coll...

  29. [37]

    Bidirectional copy-paste for semi-supervised medical image segmentation,

    Y . Bai, D. Chen, Q. Li, W. Shen, and Y . Wang, “Bidirectional copy-paste for semi-supervised medical image segmentation,” 2023. [Online]. Available: https://arxiv.org/abs/2305.00673

  30. [38]

    Caussl: Causality- inspired semi-supervised learning for medical image segmentation,

    J. Miao, C. Chen, F. Liu, H. Wei, and P.-A. Heng, “Caussl: Causality- inspired semi-supervised learning for medical image segmentation,” in 2023 IEEE/CVF International Conference on Computer Vision (ICCV) , 2023, pp. 21 369–21 380

  31. [39]

    Weight-averaged consistency targets improve semi-supervised deep learning results,

    A. Tarvainen and H. Valpola, “Weight-averaged consistency targets improve semi-supervised deep learning results,” CoRR, vol. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 12 abs/1703.01780, 2017. [Online]. Available: http://arxiv.org/abs/1703. 01780

  32. [40]

    Uncertainty- aware self-ensembling model for semi-supervised 3d left atrium segmentation,

    L. Yu, S. Wang, X. Li, C. Fu, and P. Heng, “Uncertainty- aware self-ensembling model for semi-supervised 3d left atrium segmentation,” CoRR, vol. abs/1907.07034, 2019. [Online]. Available: http://arxiv.org/abs/1907.07034

  33. [41]

    Shape-aware semi-supervised 3d semantic segmentation for medical images,

    S. Li, C. Zhang, and X. He, “Shape-aware semi-supervised 3d semantic segmentation for medical images,” CoRR, vol. abs/2007.10732, 2020. [Online]. Available: https://arxiv.org/abs/2007.10732

  34. [42]

    Semi-supervised medical image segmentation through dual-task consistency,

    X. Luo, J. Chen, T. Song, Y . Chen, G. Wang, and S. Zhang, “Semi-supervised medical image segmentation through dual-task consistency,” CoRR, vol. abs/2009.04448, 2020. [Online]. Available: https://arxiv.org/abs/2009.04448

  35. [43]

    Semi-supervised segmentation via uncertainty rectified pyramid consistency and its application to gross target volume of nasopharyngeal carcinoma,

    X. Luo, W. Liao, J. Chen, T. Song, Y . Chen, G. Wang, and S. Zhang, “Semi-supervised segmentation via uncertainty rectified pyramid consistency and its application to gross target volume of nasopharyngeal carcinoma,” CoRR, vol. abs/2012.07042, 2020. [Online]. Available: https:...

  36. [44]

    Semi-supervised semantic segmentation with cross pseudo supervision,

    X. Chen, Y . Yuan, G. Zeng, and J. Wang, “Semi-supervised semantic segmentation with cross pseudo supervision,” CoRR, vol. abs/2106.01226, 2021. [Online]. Available: https://arxiv.org/abs/2106. 01226

  37. [45]

    Enforcing mutual consistency of hard regions for semi-supervised medical image segmentation,

    Y . Wu, Z. Ge, D. Zhang, M. Xu, L. Zhang, Y . Xia, and J. Cai, “Enforcing mutual consistency of hard regions for semi-supervised medical image segmentation,” CoRR, vol. abs/2109.09960, 2021. [Online]. Available: https://arxiv.org/abs/2109.09960 VI. A PPENDIX A. Network Archite...

  38. [46]

    Similar to the encoder, the decoder of UNet also consists of 4 convolutional blocks and a projection head

    The kernel size of Conv2d is 3× 3 with 1 zero padding. Similar to the encoder, the decoder of UNet also consists of 4 convolutional blocks and a projection head. Each convo- lutional block consists of a ConvTranspose2d followed by two ” Conv2d→ BatchNorm2d→ LeakyReLU” structur...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.