Pith. sign in

REVIEW 4 major objections 2 minor 71 references

Hybrid(Transformer+CNN)-based Polyp Segmentation

T0 review · 4 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper claims a hybrid Swin Transformer-CNN network outperforms existing polyp-segmentation models, lifting recall to 0.9555 and accuracy to 0.9849 on Kvasir-SEG.

desk verdict Test-set early stopping and a missing architecture description leave the SOTA claim unsupported; the paper is a clear but overclaimed incremental baseline. read the letter →

arxiv 2508.09189 v1 pith:337YT7D4 submitted 2025-08-08 eess.IV cs.AIcs.CV

Madan Baduwal This is my paper
classification eess.IVcs.AIcs.CV
keywords colorectalcancerpolypsegmentationcolonoscopyCNNTransformerSwinhybridarchitectureKvasir-SEG
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

Colonoscopists miss a fifth or more of polyps, and segmentation models that look accurate on benchmarks can still fail on flat lesions with blurred edges. The paper argues that a hybrid network—Swin Transformer encoder plus lightweight CNN decoder—cuts those misses: on Kvasir-SEG it reports recall of 0.9555 (1.76 percentage points above DUCK-Net) and accuracy of 0.9849. The authors trace the gain to boundary-aware attention and artifact resilience, and claim the model also runs at the highest frames per second among the compared methods. If those numbers hold, the architecture would be a practical option for real-time colonoscopy assistance rather than just an offline benchmark improvement.

What carries the argument

The central object is the staged hybrid encoder-decoder: a Swin Transformer backbone emitting feature maps $F_1,\dots,F_4$ at resolutions $H/4$, $H/8$, $H/16$, $H/32$ with channel depths $C,2C,4C,8C$; a decoder $D_4,\dots,D_1$ that refines each level with a $3\times3$ conv-BN-ReLU block, bilinear 2x upsampling, and concatenative skip connections; and a final $1\times1$ convolution plus bilinear resize to full resolution. The paper names adaptive fusion modules, context-preserving skip connections, and a cross-attention bridge as the three innovations that balance transformer and CNN features. The mechanism's job is to join long-range global context with pixel-level boundary detail, which is

What would settle it

Rerun the same 900/100 Kvasir-SEG training recipe but stop learning based on a held-out validation set rather than the test set, then recompute recall and accuracy on the test images. If recall falls below DUCK-Net's 0.9379 or accuracy below 0.9842, the paper's central superiority claim fails; a quicker check is to compute the missing mDice/mIoU cells in Table 2 and see whether they also exceed DUCK-Net.

Watch

Extended reading notes

Core claim

Central claim: a hybrid (Transformer + CNN) architecture can outperform established polyp-segmentation models on the Kvasir-SEG benchmark. The paper reports recall 0.9555 and accuracy 0.9849, improving on DUCK-Net's recall (0.9379) and accuracy (0.9842), and states that this comes with the highest FPS among the models compared. The design pairs a pre-trained Swin Transformer encoder—which produces four hierarchical feature maps from $H/4$ to $H/32$ using shifted-window self-attention—with a CNN decoder that restores spatial detail through conv-BN-ReLU blocks, bilinear upsampling, and concatenative skip connections. The author's explanation is that this division of labor handles both ill-defi

Load-bearing premise

The reported recall and accuracy gains would still appear on genuinely unseen colonoscopy images, even though training was stopped by watching the Dice score on the same test set.

Editorial extensions

If this is right

  • On Kvasir-SEG, the hybrid model would detect a larger share of polyp pixels than DUCK-Net (recall 0.9555 vs 0.9379), raising the chance that small or flat lesions are visible to the endoscopist.
  • The reported frames-per-second means the accuracy gain would not necessarily block live 1080p/30fps use, making the model a candidate for real-time assistance.
  • If the multi-dataset tests on CVC-ClinicDB, ETIS, CVC-ColonDB, and Endotect are as claimed, the design transfers across different scopes, resolutions, and polyp appearances.
  • The same encoder-decoder split—transformer for context, CNN for detail—would apply to other medical segmentation problems where boundaries are weak.

Reading between the lines

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

  • The paper's own Table 2 leaves mDice and mIoU blank for the hybrid model while DUCK-Net reports 0.9051 mIoU; filling those cells with the same test set would show whether the 'superior performance' claim is purely a recall-and-accuracy result or holds across overlap metrics.
  • The abstract credits boundary-aware attention, but Section 3's equations specify only conv-BN-ReLU blocks and bilinear upsampling; an ablation that turns off the cross-attention bridge would reveal whether the recall gain actually comes from explicit boundary modeling or from the multi-scale skip connections alone.
  • A natural extension is to run the same 900/100 Kvasir-SEG split with early stopping on a validation split and to publish per-dataset results for CVC-ClinicDB, ETIS, CVC-ColonDB, and Endotect; that would separate architecture gains from test-set tuning.
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 / 2 minor

Summary. The paper proposes a hybrid Swin-Transformer/CNN architecture for polyp segmentation and reports results on Kvasir-SEG, claiming superior performance over state-of-the-art methods, including DUCK-Net. The architecture is a standard encoder–decoder with skip connections, and the reported evaluation uses a 900/100 train/test split. The central claims are that the model improves recall and accuracy over DUCK-Net and that it is robust to endoscopic artifacts and ill-defined polyp boundaries.

Significance. If the reported results were obtained under a sound protocol and the architecture matched the described innovations, the work could be of interest for real-time polyp segmentation. The paper is clearly written and the architecture is easy to follow. However, the evaluation protocol is flawed, the reported metrics do not support the superiority claim even at face value, and several abstract-level claims are not backed by experiments or by the described architecture. The current manuscript does not provide reliable evidence for its central claim.

major comments (4)
  1. [Section 4 and §5.1] The early-stopping criterion is explicitly computed on the test set: 'inspecting the Dice score on the test set after every epoch' and 'cross-validation was applied solely on the test set within the Kvasir dataset.' This means the reported test metrics are not unbiased estimates of generalization; they are selected using the same data on which they are reported. Any comparison with DUCK-Net's published numbers is therefore invalid unless DUCK-Net was selected under the same protocol. A separate validation split is required before any claim of superiority can be made.
  2. [Table 2 and §6] Even accepting the table at face value, the model is not superior: F1 is 0.9499 versus DUCK-Net's 0.9502, precision is 0.9422 versus 0.9628, and mDice/mIoU for the proposed model are missing (dash). The only claimed gains are recall and accuracy, and the recall value is inconsistent between the abstract (0.9555) and Table 2 (0.955). To support the central claim, the authors need to report the complete metric set, ideally with confidence intervals, and compare all models under the same evaluation protocol.
  3. [Abstract versus §3] The abstract attributes the performance to 'boundary-aware attention mechanisms' and robustness to 'specular highlights, motion blur, and fluid occlusions,' but the architecture described in §3.1–3.3 is a standard Swin encoder with CNN decoder blocks and concatenation skip connections; there is no boundary-aware attention module, and no artifact-robustness experiments are reported anywhere in the manuscript. These claims are therefore unsupported by the described method and experiments.
  4. [§5.1 and §7] The paper states that the model was further tested on CVC-ClinicDB, ETIS, CVC-ColonDB, and Endotect, and the conclusion claims SOTA performance across DSC, IoU, precision, recall, F2, and FPS. However, no results for any external dataset are presented, and Table 2 does not report mDice, mIoU, F2, or FPS for the proposed model. These claims should be substantiated or removed.
minor comments (2)
  1. [Throughout] There are typos and formatting issues: 'Kavasir-SEG' should be 'Kvasir-SEG', '33 convolutional' should be '3×3 convolutional', 'coscusative performance' appears to be a typo, and Figure 4's caption says 'fifty epochs' while Section 4 states training ran for 100 epochs.
  2. [References] The reference list contains entries that appear to be irrelevant to the manuscript (e.g., [4], [15], [46], [51]) and some citations are not properly matched to the text. The references should be pruned and verified.

Circularity Check

1 steps flagged · score 6.0 of 10

Test-set early stopping turns the reported recall/accuracy into fitted values, so the SOTA claim is partially circular.

  1. fitted input called prediction [Section 4 (Implementation details), early-stopping paragraph; Section 5.1; Table 2]
    "In an effort to avoid overfitting, an early stopping condition was used, inspecting the Dice score on the test set after every epoch. Training was stopped if no progress had been made in the last 37 epochs, which happened at epoch 63."

    The stopping epoch was selected by monitoring Dice on the Kvasir-SEG test set. The same test set then produced the Recall=0.9555 and Accuracy=0.9849 in Table 2, which the abstract highlights as 'improved by 1.76%... 0.07%' over DUCK-Net. Section 5.1 reinforces: 'cross-validation was applied solely on the test set within the Kvasir dataset.' Thus the headline metrics are not out-of-sample predictions; they are outcomes of a test-set selection procedure. Comparing them to DUCK-Net's independently evaluated published numbers is invalid, and the claimed gains (recall, accuracy) are precisely the metrics this protocol can inflate. The central 'superior performance' claim reduces to a test-set fit rather than an independent prediction.

full rationale

This paper has no theoretical derivation that could be self-definitional; it is an empirical architecture paper. The architecture (Swin encoder + CNN decoder with concatenation and upsampling) is presented directly, with no imported uniqueness claim and no self-citations, so patterns 3-5 do not apply. The one load-bearing circularity is in the evaluation protocol: Section 4 states that early stopping inspected the Dice score on the test set after every epoch and stopped at epoch 63, and Section 5.1 says cross-validation was applied solely on the test set within Kvasir. The same test set then supplies the Table 2 metrics, including Recall 0.9555 and Accuracy 0.9849, which the abstract promotes as SOTA improvements over DUCK-Net. Those numbers are therefore selected values, not independent generalization estimates, and the comparison to DUCK-Net's published numbers is invalid under this protocol. This is a genuine 'fitted input called prediction' pattern and it sits at the center of the paper's superiority claim, so the circularity score is 6. Separately, the abstract's claims of boundary-aware attention and artifact resilience are not supported by Section 3 or by any artifact-ablation experiment, but that is a missing-support/correctness issue, not circularity, and I do not count it in the score.

Assumptions & free parameters 1 free parameters · 2 assumptions · 0 invented entities

The model is a standard Swin Transformer encoder with a convolutional decoder. No new theoretical constants or entities are introduced. The main free choice affecting the claim is the set of hyperparameters plus the test-set early stopping rule.

free parameters (1)
  • Training hyperparameters (learning rate, weight decay, batch size, input size, early stopping patience) = 1e-4, 1e-4, 8, 352x352, 37 epochs
    Chosen by hand and not ablated; the central performance numbers depend on these choices, particularly the early-stopping rule that used test-set Dice.
assumptions (2)
  • domain assumption Kvasir-SEG 900/100 split is a valid benchmark for polyp segmentation
    The paper trains once on 900 images and reports on 100; no external validation is shown.
  • ad hoc to paper Using the test set for early stopping yields unbiased final metrics
    Section 4 states early stopping inspected Dice on the test set; this assumption is false and inflates results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hybrid(Transformer+CNN)-based Polyp Segmentation." pith.science (2026). https://pith.science/paper/337YT7D4

@misc{pith2026250809189,
  author       = {Pith},
  title        = {Pith review of: Hybrid(Transformer+CNN)-based Polyp Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/337YT7D4}},
  note         = {Machine review of arXiv:2508.09189}
}
read the original abstract

Colonoscopy is still the main method of detection and segmentation of colonic polyps, and recent advancements in deep learning networks such as U-Net, ResUNet, Swin-UNet, and PraNet have made outstanding performance in polyp segmentation. Yet, the problem is extremely challenging due to high variation in size, shape, endoscopy types, lighting, imaging protocols, and ill-defined boundaries (fluid, folds) of the polyps, rendering accurate segmentation a challenging and problematic task. To address these critical challenges in polyp segmentation, we introduce a hybrid (Transformer + CNN) model that is crafted to enhance robustness against evolving polyp characteristics. Our hybrid architecture demonstrates superior performance over existing solutions, particularly in addressing two critical challenges: (1) accurate segmentation of polyps with ill-defined margins through boundary-aware attention mechanisms, and (2) robust feature extraction in the presence of common endoscopic artifacts, including specular highlights, motion blur, and fluid occlusions. Quantitative evaluations reveal significant improvements in segmentation accuracy (Recall improved by 1.76%, i.e., 0.9555, accuracy improved by 0.07%, i.e., 0.9849) and artifact resilience compared to state-of-the-art polyp segmentation methods.

Figures

Figures reproduced from arXiv: 2508.09189 by the authors.

Figure 1
Figure 1. Hybrid (Transformer+CNN)-based Polyp Segmentation Architecture [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 3
Figure 3. Comparative visualization of polyp segmentation [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figure 4
Figure 4. Evaluation of our model in the Kvasir-SEG dataset, fifty epochs [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

71 extracted references · 67 canonical work pages

  1. [1]

    Edd2020: A comprehensive dataset for en- doscopic artifact detection

    Sharib Ali et al. Edd2020: A comprehensive dataset for en- doscopic artifact detection. Endoscopy, 52:S1–S7, 2020. 4

    work page 2020

  2. [2]

    Polypgen: A multi-center polyp detection and segmentation dataset

    Sharib Ali et al. Polypgen: A multi-center polyp detection and segmentation dataset. Nature Scientific Data, 9(1):1–12,

    work page

  3. [3]

    Sharib Ali, Feng Zhou, Barbara Braden, Adam Bailey, Suhui Yang, Guangtao Cheng, Pengyi Zhang, Xiao Li, Max Kayser, Roger Soberanis-Mukul, et al. Automatic detec- 5 Table 2: Performance Comparison of Polyp Segmentation Models in Kavasir-SEG dataset Model F1 Score mDice mIoU Precision Recall Accuracy U-Net† [47] 0.8655 – 0.7629 0.8593 0.8718 - ResUNet [68] ...

    work page 2016

  4. [4]

    Artificial intelligence in knee arthroplasty: Current concept of the available clinical applications

    Christophe Batailler, Jeremy Shatrov, Emmanuel Sappey- Marinier, Elvire Servien, Sebastien Parratte, and S ´ebastien Lustig. Artificial intelligence in knee arthroplasty: Current concept of the available clinical applications. Arthroplasty, 4:1–16, 2022

    work page 2022

  5. [5]

    To- wards automatic polyp detection with a polyp appearance model

    Jorge Bernal, F Javier S ´anchez, Gloria Fern´andez-Esparrach, Debora Gil, Cristina Rodr´ıguez, and Fernando Vilarino. To- wards automatic polyp detection with a polyp appearance model. Pattern Recognition, 48(11):3166–3182, 2015. 4

    work page 2015

  6. [6]

    Sanchez, G

    Jorge Bernal, Francisco J. Sanchez, G. Fernandez-Esparrach, D. Gil, C. Rodrıguez, and F. Vilarino. Wm-dova maps for accurate polyp highlighting in colonoscopy: Validation vs. saliency maps from physicians. Computers in Medical Imag- ing and Graphics, 43:99–111, 2015

    work page 2015

  7. [7]

    Wm- dova maps for accurate polyp highlighting in colonoscopy

    Jorge Bernal, Javier S ´anchez, and Fernando Vilarino. Wm- dova maps for accurate polyp highlighting in colonoscopy. Medical Image Analysis, 17(7):703–718, 2013. 4

    work page 2013

  8. [8]

    Comparative validation of polyp detection methods in video colonoscopy

    Jorge Bernal, Nima Tajkbaksh, Francisco Javier S ´anchez, Bogdan J Matuszewski, Hao Chen, Lequan Yu, Quentin Angermann, Olivier Romain, Bjørn Rustad, Ilangko Balas- ingham, et al. Comparative validation of polyp detection methods in video colonoscopy. IEEE Transactions on Medi- cal Imaging, 36(2):496–507, 2017. 4

    work page 2017

Show all 71 references
  1. [9]

    Un- derstanding robustness of transformers for image classifica- tion

    Srinadh Bhojanapalli, Ayan Chakrabarti, Daniel Glasner, Daliang Li, Thomas Unterthiner, and Andreas Veit. Un- derstanding robustness of transformers for image classifica- tion. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 10211–10221, 2021

    2021

  2. [10]

    Hyperkvasir: A comprehensive multi- class image and video dataset for gastrointestinal endoscopy

    Rune Borgli et al. Hyperkvasir: A comprehensive multi- class image and video dataset for gastrointestinal endoscopy. Scientific Data, 7(1):1–14, 2020. 4

    2020

  3. [11]

    Swin-unet: Unet-like pure transformer for medical image segmentation

    Hu Cao, Yueyue Wang, Joy Chen, Dongsheng Jiang, Xi- aopeng Zhang, Qi Tian, and Manning Wang. Swin-unet: Unet-like pure transformer for medical image segmentation. arXiv preprint arXiv:2105.05537, 2021. 3

    2021 arXiv

  4. [12]

    Yuille, and Yuyin Zhou

    Jieneng Chen, Yongyi Lu, Qihang Yu, Xiangde Luo, Ehsan Adeli, Yan Wang, Le Lu, Alan L. Yuille, and Yuyin Zhou. Transunet: Transformers make strong encoders for medi- cal image segmentation. arXiv preprint arXiv:2102.04306,

    arXiv

  5. [13]

    Polyp-pvt: Polyp segmentation with pyramid vision transformers

    Bing Dong, Wenhai Wang, Deng-Ping Fan, Jian Li, Huazhu Fu, and Ling Shao. Polyp-pvt: Polyp segmentation with pyramid vision transformers. ArXiv preprint, 2021. 3

    2021

  6. [14]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, et al. An image is worth 16x16 words: Transformers for image recognition at scale. ArXiv preprint, 2021. 3

    2021

  7. [15]

    Espinosa, Gaston A

    Marco G. Espinosa, Gaston A. Otarola, Jerry C. Hu, and Kyr- iacos A. Athanasiou. Cartilage assessment requires a surface characterization protocol: Roughness, friction, and function. Tissue Engineering Part C: Methods, 27:276–286, 2021

    2021

  8. [16]

    Sun-seg: A large-scale dataset for sur- gical scene segmentation

    Deng-Ping Fan et al. Sun-seg: A large-scale dataset for sur- gical scene segmentation. IEEE Transactions on Medical Imaging, 41(3):786–796, 2022. 4

    2022

  9. [17]

    Pranet: Parallel reverse attention network for polyp segmentation

    Deng-Ping Fan, Ge-Peng Ji, Tao Zhou, Geng Chen, Huazhu Fu, Jianbing Shen, and Ling Shao. Pranet: Parallel reverse attention network for polyp segmentation. In International Conference on Medical Image Computing and Computer- Assisted Intervention, pages 263–273. Springer, 2020. 2, 6

    2020

  10. [18]

    Roth, and Daguang Xu

    Ali Hatamizadeh, Vishwesh Nath, Yucheng Tang, Dong Yang, Holger R. Roth, and Daguang Xu. Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images. In International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI) , ...

    2022

  11. [19]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 770–778, 2016. 2

    2016

  12. [20]

    Kvasir-capsule: A video capsule en- doscopy dataset

    Steven A Hicks et al. Kvasir-capsule: A video capsule en- doscopy dataset. Nature Scientific Reports, 11:15671, 2021. 4

    2021

  13. [21]

    Piccolo dataset for endoscopic polyp segmentation

    Alejandro Jaqu ´ez et al. Piccolo dataset for endoscopic polyp segmentation. Medical Physics, 47:e123–e124, 2020. 4

    2020

  14. [22]

    Real- time polyp detection, localization and segmentation in colonoscopy using deep learning

    Debesh Jha, Safdar Ali, Nikhil Kumar Tomar, et al. Real- time polyp detection, localization and segmentation in colonoscopy using deep learning. IEEE Access, 9:40496– 40510, 2021. 2

    2021

  15. [23]

    Johansen, Dag Johansen, Jens Rittscher, Michael A

    Debesh Jha, Sharib Ali, Naman Kumar Tomar, H ˚avard D. Johansen, Dag Johansen, Jens Rittscher, Michael A. Riegler, and P˚al Halvorsen. Doubleu-net: A deep convolutional neu- ral network for medical image segmentation. 2020 IEEE 33rd International Symposium on Computer-Based Me...

    2020

  16. [24]

    Smedsrud, Daniel Johansen, et al

    Debesh Jha, Pia H. Smedsrud, Daniel Johansen, et al. A comprehensive study on colorectal polyp segmentation with resunet++, conditional random field and test-time augmenta- tion. IEEE Journal of Biomedical and Health Informatics , 25:2029–2040, 2021. 2

  17. [25]

    Smedsrud, Michael A

    Debesh Jha, Pia H. Smedsrud, Michael A. Riegler, et al. Kvasir-seg: a segmented polyp dataset. In Interna- tional Conference on Multimedia Modeling, pages 451–462. Springer, Cham, 2020. 2

    2020

  18. [26]

    Kvasir-seg: A segmented polyp dataset

    Debesh Jha, Pia H Smedsrud, Michael A Riegler, P ˚al Halvorsen, Thomas de Lange, Dag Johansen, and H ˚avard D Johansen. Kvasir-seg: A segmented polyp dataset. arXiv preprint arXiv:1911.07069, 2019. 4

    1911 arXiv

  19. [27]

    Smedsrud, Michael A

    Debesh Jha, Pia H. Smedsrud, Michael A. Riegler, Daniel T. Johansen de Lange, P˚al Halvorsen, and H˚avard D. Johansen. Resunet++: an advanced architecture for medical image seg- mentation. In Proceedings of IEEE International Symposium on Multimedia (ISM), pages 225–255, 2019. 2, 6

    2019

  20. [28]

    Nanonet: real-time polyp segmentation in video capsule endoscopy and colonoscopy

    Debesh Jha, Nikhil Kumar Tomar, Safdar Ali, et al. Nanonet: real-time polyp segmentation in video capsule endoscopy and colonoscopy. InProceedings of IEEE International Sym- posium on Computer-Based Medical Systems (CBMS), pages 37–43, 2021. 2, 3

    2021

  21. [29]

    Deep learning-based detection of polyps in colonoscopy

    Tae Kyoung Kim, Chan Ho Park, Chang Min Lee, Min Ju Kang, and Hyun Gun Kim. Deep learning-based detection of polyps in colonoscopy. Scientific Reports, 10(1):1–9, 2020. 4

    2020

  22. [30]

    Li-segpnet: A lightweight pyramid network for real-time polyp segmentation

    Yuan Li, Hao Chen, Manning Wang, and Xiaolong Zhang. Li-segpnet: A lightweight pyramid network for real-time polyp segmentation. IEEE Transactions on Medical Imag- ing, 41(5):1124–1135, 2022. 6

    2022

  23. [31]

    Sun: A large-scale dataset for surgical understanding

    Daochang Liu, Ziyu Jiang, Yizhou Wang, Qiang Wang, Fei Wang, Ziyu Li, and Changhu Wang. Sun: A large-scale dataset for surgical understanding. IEEE Transactions on Medical Robotics and Bionics, 2(1):41–48, 2020. 4

    2020

  24. [32]

    Cp-child: A pediatric colon polyp dataset

    Jiang Liu et al. Cp-child: A pediatric colon polyp dataset. Scientific Data, 7(1):1–8, 2020. 4

    2020

  25. [33]

    Ddanet: Dual decoder attention network for automatic polyp segmentation

    Jiang Liu, Tengfei Song, Meng Li, Linlin Qiao, Yitian Zhao, and Yunde Jia. Ddanet: Dual decoder attention network for automatic polyp segmentation. Pattern Recognition , 122:108318, 2022. 2

    2022

  26. [34]

    Ld polyp video: A large-scale colonoscopy video dataset

    Xiaoyu Liu et al. Ld polyp video: A large-scale colonoscopy video dataset. Medical Image Analysis, 70:101987, 2021. 4

    2021

  27. [35]

    Swin trans- former: Hierarchical vision transformer using shifted win- dows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin trans- former: Hierarchical vision transformer using shifted win- dows. ICCV, 2021

    2021

  28. [36]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 3431–3440, 2015. 2

    2015

  29. [37]

    Decoupled weight de- cay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight de- cay regularization. In International Conference on Learning Representations (ICLR), 2019

    2019

  30. [38]

    Mamonov, Isabel N

    Alexander V . Mamonov, Isabel N. Figueiredo, Pedro N. Figueiredo, and Yu-Hsiang R. Tsai. Automated polyp de- tection in colon capsule endoscopy. IEEE Transactions on Medical Imaging, 33:1488–1502, 2014. 2

    2014

  31. [39]

    Matsuda, A

    T. Matsuda, A. Ono, M. Sekiguchi, T. Fujii, and Y . Saito. Advances in image enhancement in colonoscopy for detec- tion of adenomas. Nature Reviews Gastroenterology & Hep- atology, 14:305–314, 2017. 1

    2017

  32. [40]

    Understanding your diagno- sis: Colonoscopy

    The American Cancer Society Medical and Edito- rial Content Team. Understanding your diagno- sis: Colonoscopy. https://www.cancer.org/ treatment/understanding-your-diagnosis/ tests/endoscopy/colonoscopy.html, n.d. Accessed: YYYY-MM-DD. 1

  33. [41]

    A deep learning method for early detection of dia- betic foot using decision fusion and thermal images

    Khairul Munadi, Khairunnisa Saddami, Masayu Oktiana, et al. A deep learning method for early detection of dia- betic foot using decision fusion and thermal images. Applied Sciences, 12:7524–7545, 2022

    2022

  34. [42]

    Cancer stat facts: Colorectal cancer

    NIH. Cancer stat facts: Colorectal cancer. http: //www.seer.cancer.gov/statfacts/html/ colorect.html, n.d. Accessed: YYYY-MM-DD. 1

  35. [43]

    Uacanet: Unified adaptive context-aware network for polyp segmentation

    Seung-Jun Oh, Hyun Kim, Sang-Hoon Oh, and Seung-Won Lee. Uacanet: Unified adaptive context-aware network for polyp segmentation. In Proceedings of the 29th ACM In- ternational Conference on Multimedia , pages 1762–1770. ACM, 2021. 2

    2021

  36. [44]

    Deep learning and hand-crafted fea- tures for automatic polyp detection

    Konstantin Pogorelov, Klaus R Randel, Carsten Griwodz, Sigrun L Eskeland, Thomas de Lange, Dag Johansen, Con- cetto Spampinato, Duc-Tien Dang-Nguyen, Mathias Lux, Peter T Schmidt, et al. Deep learning and hand-crafted fea- tures for automatic polyp detection. In IEEE Internati...

  37. [45]

    Nbi and wl ucdb databases for computer-assisted detection of ulcerative lesions

    Sara Reis, Guilherme Macedo, Cl ´audia Bahia, and Miguel T Coimbra. Nbi and wl ucdb databases for computer-assisted detection of ulcerative lesions. IEEE Access , 7:74293– 74302, 2019. 4

    2019

  38. [46]

    Rodr ´ıguez-Merch´an and Pilar G ´omez-Cardero

    Eduardo C. Rodr ´ıguez-Merch´an and Pilar G ´omez-Cardero. The outerbridge classification predicts the need for patellar resurfacing in tka. Clinical Orthopaedics and Related Re- search, 468:1254–1257, 2010

    2010

  39. [47]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Inter- vention (MICCAI), volume 9351, pages 234–241, 2015. 2, 6

    2015

  40. [48]

    Toward embedded detection of polyps in wce images for early diagnosis of colorectal can- cer

    Jo ˜ao Silva, Aymeric Histace, Olivier Romain, Xavier Dray, and Bertrand Granado. Toward embedded detection of polyps in wce images for early diagnosis of colorectal can- cer. In IEEE International Conference on Robotics and Biomimetics, pages 1790–1795, 2014. 4

    2014

  41. [49]

    Toward embedded detection of polyps in wce images for early diagnosis of colorectal can- cer

    Jorge Silva, Aymeric Histace, Olivier Romain, Xavier Dray, and Bertrand Granado. Toward embedded detection of polyps in wce images for early diagnosis of colorectal can- cer. International Journal of Computer Assisted Radiology and Surgery, 9:283–293, 2014

    2014

  42. [50]

    Very deep con- volutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep con- volutional networks for large-scale image recognition. In In- ternational Conference on Learning Representations (ICLR),

  43. [51]

    Colleen Slattery and Charles Y . Kweon. Classifications in brief: Outerbridge classification of chondral lesions.Clinical Orthopaedics and Related Research, 476:2101–2104, 2018

    2018

  44. [52]

    Kvasir-capsule, a video capsule endoscopy dataset

    Pia H Smedsrud, Vajira Thambawita, Steven A Hicks, H˚avard L Gjestang, Oda O Nedrejord, Espen Næss, Rune Borgli, Debesh Jha, Torunn Berstad, Sigrun L Eskeland, et al. Kvasir-capsule, a video capsule endoscopy dataset. arXiv preprint arXiv:1907.05719, 2019. 4

    1907 arXiv

  45. [53]

    Gurudu, and Jianming Liang

    Nima Tajbakhsh, Suryakanth R. Gurudu, and Jianming Liang. Automated polyp detection in colonoscopy videos using shape and context information. IEEE Transactions on Medical Imaging, 35:630–644, 2015

    2015

  46. [54]

    Automated polyp detection in colonoscopy videos using shape and context information

    Nima Tajbakhsh, Suryakanth R Gurudu, and Jianming Liang. Automated polyp detection in colonoscopy videos using shape and context information. IEEE Transactions on Medical Imaging, 35(2):630–644, 2016. 4

    2016

  47. [55]

    Vajira Thambawita, Steven Hicks, P ˚al Halvorsen, and Michael A. Riegler. Pyramid-focus-augmentation: Medical image segmentation with step-wise focus. ArXiv preprint,

  48. [56]

    Atten- tion is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, et al. Atten- tion is all you need. In Advances in Neural Information Pro- cessing Systems (NeurIPS), pages 6000–6010, 2017. 3

    2017

  49. [57]

    Sanchez, et al

    David Vazquez, Jorge Bernal, Francisco J. Sanchez, et al. A benchmark for endoluminal scene segmentation of colonoscopy images. Journal of Healthcare Engineering , 2017:1–9, 2017

    2017

  50. [58]

    Vezakis, Konstantinos Georgas, Dimitrios Fo- tiadis, and George K

    Ioannis A. Vezakis, Konstantinos Georgas, Dimitrios Fo- tiadis, and George K. Matsopoulos. Effisegnet: Gas- trointestinal polyp segmentation through a pre-trained efficientnet-based network with a simplified decoder, 2024

    2024

  51. [59]

    Duck-net: Dense u-shaped convolutional ker- nel network for polyp segmentation

    Cheng Wang, Zheng Li, Xiaolong Zhang, Minghui Qiu, and Guang Yang. Duck-net: Dense u-shaped convolutional ker- nel network for polyp segmentation. IEEE Transactions on Medical Imaging, 42(4):1029–1040, 2023. 6

    2023

  52. [60]

    Nanonet: Real-time polyp segmentation with ultra-lightweight models for edge devices

    Cheng Wang, Zheng Li, Xiaolong Zhang, Minghui Qiu, and Guang Yang. Nanonet: Real-time polyp segmentation with ultra-lightweight models for edge devices. In IEEE In- ternational Conference on Bioinformatics and Biomedicine (BIBM), pages 1234–1241. IEEE, 2023. 3

    2023

  53. [61]

    Pvt v2: Improved baselines with pyramid vision transformer

    Wenhai Wang, Enze Xie, Xiang Li, et al. Pvt v2: Improved baselines with pyramid vision transformer. Computer Vision and Image Understanding (CVIU), 8:415–424, 2021. 3

    2021

  54. [62]

    A versatile back- bone for dense prediction without convolutions

    Wenhai Wang, Enze Xie, Xiang Li, et al. A versatile back- bone for dense prediction without convolutions. In Proceed- ings of the IEEE/CVF International Conference on Com- puter Vision (ICCV), pages 548–558, 2021. 3

    2021

  55. [63]

    Msrfe-net: Multi-scale residual feature enhancement network for polyp segmentation

    Yujia Wang, Yu Zhang, Jindong Tian, Cheng Zhong, Yong- dong Zhang, and Yiyan Shi. Msrfe-net: Multi-scale residual feature enhancement network for polyp segmentation. IEEE Transactions on Medical Imaging, 41(12):3721–3734, 2022. 2

    2022

  56. [64]

    Williams, Domenico Borroni, Renqiang Liu, et al

    Ben M. Williams, Domenico Borroni, Renqiang Liu, et al. An artificial intelligence-based deep learning algorithm for the diagnosis of diabetic neuropathy using corneal confo- cal microscopy: A development and validation study. Di- abetologia, 63:419–430, 2020

    2020

  57. [65]

    Alvarez, and Ping Luo

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, and Ping Luo. Segformer: Simple and effi- cient design for semantic segmentation with transformers. In Proceedings of the 35th Conference on Neural Information Processing Systems (NeurIPS), 2021

    2021

  58. [66]

    Colonformer: An efficient transformer based method for colon polyp segmentation

    Wei Zhang, Qiang Li, Zhou Yu, Yang Chen, and Jingdong Wang. Colonformer: An efficient transformer based method for colon polyp segmentation. IEEE Journal of Biomedical and Health Informatics, 26(8):3970–3980, 2022. 6

    2022

  59. [67]

    Fcb- former: A fully convolutional bridge transformer with pyra- mid squeeze-excitation for colonoscopy segmentation

    Yu Zhang, Jiang Liu, Qinghua Hu, and Manning Wang. Fcb- former: A fully convolutional bridge transformer with pyra- mid squeeze-excitation for colonoscopy segmentation. In Medical Image Computing and Computer Assisted Interven- tion (MICCAI), pages 352–363. Springer, 2023. 3

    2023

  60. [68]

    Road extraction by deep residual u-net

    Zhengxin Zhang, Qingjie Liu, and Yunhong Wang. Road extraction by deep residual u-net. IEEE Geoscience and Re- mote Sensing Letters, 15(5):749–753, 2018. 2, 6

    2018

  61. [69]

    Road ex- traction by deep residual u-net

    Zizhao Zhang, Qingjie Liu, and Yunhong Wang. Road ex- traction by deep residual u-net. IEEE Geoscience and Re- mote Sensing Letters, 15:749–753, 2018

    2018

  62. [70]

    Unet++: A nested u-net architecture for medical image segmentation

    Zongwei Zhou, Md Mahfuzur Rahman Siddiquee, Nima Tajbakhsh, and Jianming Liang. Unet++: A nested u-net architecture for medical image segmentation. arXiv preprint arXiv:1807.10165, 2018. 2 8

    2018 arXiv

  63. [2015]

    https://arxiv.org/abs/1409.1556. 2 7

    arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.