Paper Info Reviews Author Feedback Meta-Review Back to top
List of Papers Browse by Subject Areas Author List

Abstract

Accurate polyp segmentation methods are essential for colon polyp screening and colorectal cancer diagnosis. However, polyp segmentation faces the following challenges: (1) Small-sized polyps are easily lost during the identification process. (2) The boundaries separating the polyp from its surroundings are fuzzy. (3) Additional distracting information is introduced during the colonoscopy procedure, resulting in noise in the colonoscopy image and influencing the segmentation outcomes. To cope with these three challenges, a method for colon polyp segmentation based on local feature supplementation and shallow feature supplementation (LSSNet) is proposed by incorporating feature supplementation structures in the encoder-decoder structure. The multiscale feature extraction (MFE) module is designed to extract local features, the interlayer attention fusion (IAF) module is designed to fuse supplementary features with the current layer features, and the semantic gap reduction (SGR) module is designed to reduce the semantic gaps between the layers, which together form the local feature supplementation structure. The shallow feature supplementation (SFS) module is designed to supplement the features in the fuzzy areas. Based on these four modules LSSNet is proposed. LSSNet is evaluated on five datasets: ClinicDB, KvasirSEG, ETIS, ColonDB, and EndoScene. The results show that mDice scores are improved by 1.33%, 0.74%, 2.65%, 1.08%, and 0.62% respectively over the compared state-of-the-art methods. The codes are available at https://github.com/heyeying/LSSNet.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2024/paper/1926_paper.pdf

SharedIt Link: https://rdcu.be/dV5Ek

SpringerLink (DOI): https://doi.org/10.1007/978-3-031-72104-5_43

Supplementary Material: N/A

Link to the Code Repository

https://github.com/heyeying/LSSNet

Link to the Dataset(s)

https://polyp.grand-challenge.org/CVCClinicDB/ https://datasets.simula.no/kvasir-seg/

BibTex

@InProceedings{Wan_LSSNet_MICCAI2024,
        author = { Wang, Wei and Sun, Huiying and Wang, Xin},
        title = { { LSSNet: A Method for Colon Polyp Segmentation Based on Local Feature Supplementation and Shallow Feature Supplementation } },
        booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024},
        year = {2024},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 15007},
        month = {October},
        page = {446 -- 456}
}

Reviews

Review #1

  • Please describe the contribution of the paper

    The paper proposes a network structure for colon polyp segmentation. The experimental results show that the structure works well on several datasets, compared with involved methods.

  • Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    Strong results show that the proposed structure helps improve the segmentation results.

  • Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
    1. lacking enough motivation to explain the intrinsic characteristic of each designed module;
    2. I wonder about the segmentation results of nnUNet and SwinUNet, which are strong baselines in various segmentation tasks.
  • Please rate the clarity and organization of this paper

    Very Good

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission has provided an anonymized link to the source code, dataset, or any other dependencies.

  • Do you have any additional comments regarding the paper’s reproducibility?

    No.

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html
    1. improving the motivation part and introducing why the designed modules help improve the performance, not just using ablation experiments.
    2. please involve several strong baselines in medical segmentation, such as nnUNet and SwinUNet, not just using UNet as the baseline.
  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making

    Weak Reject — could be rejected, dependent on rebuttal (3)

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
    1. lacking enough motivation introduction;
    2. baselines are not strong.
  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

    N/A

  • [Post rebuttal] Please justify your decision

    N/A



Review #2

  • Please describe the contribution of the paper

    A network is proposed for Colon Polyp Segmentation and the accuracy is high.

  • Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    Many structure is used in this paper, but nothing is new. However, those modules result in the good performance.

  • Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.

    No description in the title of Fg.1. Nothing is new in this paper but the performance is high with out a heavy backbone.

  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission does not mention open access to source code or data but provides a clear and detailed description of the algorithm to ensure reproducibility.

  • Do you have any additional comments regarding the paper’s reproducibility?

    I’m not sure whether it’s able to reproduce or not.

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html

    This paper seems more like a technique report. It is better to rewrite.

  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making

    Weak Accept — could be accepted, dependent on rebuttal (4)

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    This paper is not clear and should be rewritten. It’s a fusion of a set of recent modules, but nothing is explained. However, as a medicine paper, the performance is more important for practical usage.

  • Reviewer confidence

    Somewhat confident (2)

  • [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

    N/A

  • [Post rebuttal] Please justify your decision

    N/A



Review #3

  • Please describe the contribution of the paper

    This paper introduces a local feature supplementation structure, which includes the MFE module for extracting local features, the IAF module for feature fusion, and the SGR module for reducing interlayer semantic gaps. Additionally, the SFS module is proposed to complement features in the fuzzy regions. LSSNet is evaluated on multiple polyp segmentation datasets, and compared to other state-of-the-art methods, it achieves improvements in mDice scores of 1.33%, 0.74%, 2.65%, 1.08%, and 0.62% respectively.

  • Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    The paper introduces two structures, namely the local feature supplementation structure and the shallow feature supplementation structure, which enhance the accuracy of colon polyp segmentation by capturing relevant information and improving the model’s feature learning capabilities. From both quantitative and qualitative experimental results, it is evident that the proposed network effectively addresses the issue of inaccurate segmentation caused by small polyp size. Additionally, the logical presentation and graphical illustrations in the paper are clear and commendable.

  • Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.

    The main weakness of the paper lies in the insufficient experimental section. It lacks comprehensive experiments to evaluate the generalization performance of the proposed network on different datasets, such as incorporating generalization performance experiments similar to those conducted in other polyp segmentation works like PraNet. Furthermore, the comparative experimental analysis is limited, as it fails to include a comparison with advanced methods like Poly-PVT and SSFormer, which have demonstrated excellent performance in polyp segmentation tasks.

  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission does not mention open access to source code or data but provides a clear and detailed description of the algorithm to ensure reproducibility.

  • Do you have any additional comments regarding the paper’s reproducibility?

    The provided source code link in the abstract by the authors does not contain actual source code.

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html

    The experimental section in the paper is relatively limited, lacking important experiments on the generalization performance in the context of colon polyp segmentation. Furthermore, the comparative analysis is lacking in terms of the number of methods compared on each dataset. Additionally, there is a missing table caption in the ablation experiments section. It is recommended that the authors supplement the paper with additional experiments, particularly focusing on the generalization performance, and conduct a thorough review of the manuscript to address these issues. Finally, does the order and placement of these modules affect the performance of the model? The motivation behind proposing these modules is reasonable, but relying solely on current ablative experiments may not fully demonstrate it.

  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making

    Weak Accept — could be accepted, dependent on rebuttal (4)

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    The fluency of this paper is good, but it lacks some necessary experimental evidence. I believe that addressing the shortcomings of the paper will greatly enhance its quality and impact. Additionally, providing more detailed information on methods and analysis, as well as including the source code, will improve the reproducibility of the research.

  • Reviewer confidence

    Very confident (4)

  • [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

    N/A

  • [Post rebuttal] Please justify your decision

    N/A



Review #4

  • Please describe the contribution of the paper

    his paper proposes an innovative colon polyp segmentation method, effectively addressing issues such as the easy loss of small polyps, fuzzy boundaries between polyps and their surroundings, and noise interference during colonoscopy procedures. The method demonstrates significant performance improvements over existing state-of-the-art methods across multiple datasets.

  • Please list the main strengths of the paper; you should write about a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    1) Innovation: The LSSNet method introduced in this paper has clear innovative points in the field of colon polyp segmentation, especially in the design of feature supplementation structures. 2) Effectiveness: The method’s effectiveness in improving segmentation accuracy is demonstrated through extensive experiments across five public datasets. 3) Detailed Experimental Analysis: The paper provides thorough experimental results and comparisons with existing technologies, proving the method’s superiority.

  • Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.

    1) The paper may not sufficiently discuss the model’s generalization ability and performance on different types of colonoscopy images. 2) Discussions on model complexity and computational efficiency are not enough, which could affect the feasibility of practical applications.

  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission has provided an anonymized link to the source code, dataset, or any other dependencies.

  • Do you have any additional comments regarding the paper’s reproducibility?

    N/A

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review. Pay specific attention to the different assessment criteria for the different paper categories (MIC, CAI, Clinical Translation of Methodology, Health Equity): https://conferences.miccai.org/2024/en/REVIEWER-GUIDELINES.html

    The paper proposes a promising new method in the field of colon polyp segmentation. Despite some limitations, these issues are expected to be resolved with further research and improvement. The authors are encouraged to discuss the model’s generalization ability and application prospects in different scenarios in the final version in more detail.

  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making

    Weak Accept — could be accepted, dependent on rebuttal (4)

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    he paper is technically innovative and has proven its effectiveness through experiments. Although there are some weaknesses, these do not detract from the paper’s overall quality and value for acceptance.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

    Weak Accept — could be accepted, dependent on rebuttal (4)

  • [Post rebuttal] Please justify your decision

    Thanks for the authors’ rebuttal. After reading the other reviews, my opinion on this paper is unchanged.



Author Feedback

We sincerely appreciate the diligent and professional work of the reviewers, which has given us the opportunity to provide our rebuttal.

With regard to the reproducibility of the paper (R3, R4), We have made the source code accessible and intend to include additional content, which could not be accommodated in the paper due to space constraints, in future research. It is important to note that there was an issue with the anonymous URL provided for the code in the paper, and the “LSSNet-FE31” in the URL needed to be modified to “LSSNet-8C86”.

About the motivation of the study (R3, R4, R6), we want to design a model that can balance the segmentation accuracy and network complexity and prioritize the segmentation accuracy. We would like to supplement local features and polyp boundary features in the network to improve the segmentation accuracy. Therefore, we propose a Local Feature Supplementation Structure consisting of SGR, IAF, and MFE modules and a Shallow Feature Supplementation Structure consisting of the SFS module and the Convolutional branch. In terms of the order and positioning of the modules (R4), the role of the Local Feature Supplementation Structure determines the order and position of its constituent modules SGR, IAF, and MFE. SGR is responsible for receiving the local features supplemented by the previous layers and reducing the inter-layer semantic gap. IAF fuses the features of the current layer with those of the previous layers. MFE performs the extraction of multiscale features from the fused features, so these three modules are progressive. The SFS module requires the prediction map to calculate the fuzzy area, and its position is also fixed. We will further clarify the research motivation in the paper.

The generalization performance of the network (R4, R5) may not be described clearly enough in the paper, and the generalization ability experiments are consistent with PraNet. Three invisible datasets, ETIS, ColonDB, and EndoScene are used to verify the generalization ability of the network. From Table 1 in the paper, LSSNet improves the mDice on these three invisible datasets by 2.65%, 1.08%, and 0.62%, respectively. In future research, cross-validation will be added to further validate the generalization performance of the network. About the comparison experiments (R4, R6), we can only illustrate them under limited conditions as we are not allowed to add new experimental results. Due to the proprietary characteristics of polyps, it is fairer to compare polyp segmentation-specific networks rather than medical segmentation-generic networks (e.g., nnUNet and SwinUNet). Moreover, the compared polyp segmentation networks all outperform their compared SOTA methods, which makes the experimental results more convincing. U-Net is compared because LSSNet is an improvement of U-Net. We cite the experimental results of Polyp-PVT and SSFormerPVT from the paper of the comparison network CASCADE. The mDice in EndoScene, ClinicDB, Kvasir, ColonDB, and ETIS datasets are lower than CASCADE in both Polyp-PVT (88.71%, 93.08%, 91.23%, 80.75%, 78.67%) and SSFormerPVT (89.46%, 92.88%, 91.11%, 79.34%, 78.03%), and of course, lower than our LSSNet, so we do not compare these two networks. In terms of network efficiency (R5), while the Params and FLOPs of LSSNet have increased by 1.687M and 2.286G respectively compared to CASCADE, the FLOPs remain lower than the average of all the networks under comparison. The performance of LSSNet has been significantly enhanced, despite a slight increase in parameter and computational costs, making it highly valuable for practical applications.

Due to the length limitations of the paper, certain descriptions may not have been sufficiently clear (R3), and there were some detailed errors (R3, R4). We carefully reviewed the paper multiple times, rephrased unclear sections, and rectified detail inaccuracies.



Meta-Review

Meta-review #1

  • After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.

    Accept

  • Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’

    The paper proposes a Colon Polyp Segmentation network that demonstrates high accuracy. It introduces several innovative modules and achieves improvements over state-of-the-art methods across multiple datasets. The major concerns raised by reviewers were comprehensively addressed during the rebuttal. After thorough examination of the paper and all reviewers’ comments, I recommend accepting this paper.

  • What is the rank of this paper among all your rebuttal papers? Use a number between 1/n (best paper in your stack) and n/n (worst paper in your stack of n papers). If this paper is among the bottom 30% of your stack, feel free to use NR (not ranked).

    The paper proposes a Colon Polyp Segmentation network that demonstrates high accuracy. It introduces several innovative modules and achieves improvements over state-of-the-art methods across multiple datasets. The major concerns raised by reviewers were comprehensively addressed during the rebuttal. After thorough examination of the paper and all reviewers’ comments, I recommend accepting this paper.



Meta-review #2

  • After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.

    Accept

  • Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’

    rebuttal period was good, the paper is positive, overall suggestions by reviewers (avg) is weak accept+

  • What is the rank of this paper among all your rebuttal papers? Use a number between 1/n (best paper in your stack) and n/n (worst paper in your stack of n papers). If this paper is among the bottom 30% of your stack, feel free to use NR (not ranked).

    rebuttal period was good, the paper is positive, overall suggestions by reviewers (avg) is weak accept+



back to top