Paper Info Reviews Meta-Review Author Feedback Post-rebuttal Meta-Reviews Back to top
List of Papers By topics Author List
Paper Info Reviews Meta-review Author Feedback Post-Rebuttal Meta-reviews

Authors

Kit Mills Bransby, Greg Slabaugh, Christos Bourantas, Qianni Zhang

Abstract

We present a novel methodology that combines graph and dense segmentation techniques by jointly learning both point and pixel contour representations, thereby leveraging the benefits of each approach. This addresses deficiencies in typical graph segmentation methods where misaligned objectives restrict the network from learning discriminative vertex and contour features. Our joint learning strategy allows for rich and diverse semantic features to be encoded, while alleviating common contour stability issues in dense-based approaches, where pixel-level ob- jectives can lead to anatomically implausible topologies. In addition, we identify scenarios where correct predictions that fall on the contour boundary are penalised and address this with a novel hybrid contour distance loss. Our approach is validated on several Chest X-ray datasets, demonstrating clear improvements in segmentation stability and accu- racy against a variety of dense- and point-based methods. Our source code is freely available at: www.github.com/kitbransby/Joint_Graph_Segmentation

Link to paper

DOI: https://doi.org/10.1007/978-3-031-43898-1_50

SharedIt: https://rdcu.be/dnwBK

Link to the code repository

https://github.com/kitbransby/Joint_Graph_Segmentation

Link to the dataset(s)

JSRT: http://db.jsrt.or.jp/eng.php

Padchest: https://bimcv.cipf.es/bimcv-projects/padchest

Montgomery: https://data.lhncbc.nlm.nih.gov/public/Tuberculosis-Chest-X-ray-Datasets/Montgomery-County-CXR-Set/MontgomerySet/NLM-MontgomeryCXRSet-ReadMe.pdf

Shenzen: https://data.lhncbc.nlm.nih.gov/public/Tuberculosis-Chest-X-ray-Datasets/Shenzhen-Hospital-CXR-Set/index.html https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4256233/#

Reviews

Review #1

  • Please describe the contribution of the paper

    This paper develops a method with Joint Dense-Point Representation for Contour-Aware Graph Segmentation.

  • 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. clear written.
    2. simulation results are sufficient.
  • 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. how to utilize GCN in the proposed work is not clear.
    2. ablation study is missing.
  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance

    The paper has no problem on reproducibility.

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html
    1. how to utilize GCN in the proposed work is not clear.
    2. ablation study is missing.
  • Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making

    5

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
    1. how to utilize GCN in the proposed work is not clear.
    2. ablation study is missing.
  • Reviewer confidence

    Confident but not absolutely certain

  • [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

    This article proposes a novel segmentation architecture that uses the joint dense point representation method to improve accuracy and reduce topological errors.

  • 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.The method proposed in this article can leverage the advantages of both dense-based and point-based algorithms to improve segmentation accuracy. 2.The article proposes a new contour-aware loss function to address the limitations of segmentation methods using point-by-point distance. 3.The effectiveness and stability of the proposed method are verified through extensive experiments.

  • 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.This method requires the simultaneous use of dense-based and point-based algorithms during the segmentation process, which may increase computational costs. 2.The experiments in this article are only validated on chest X-ray datasets and need to be verified on other domains and datasets to demonstrate the method’s universality and scalability.

    1. In the comparison experiments, the authors did not compare their method with the latest and best-performing algorithms, but only compared it with some existing methods, which may make it difficult for readers to judge the performance of the authors’ method relative to the latest methods.
    2. The method is relatively complex. It requires considering both dense-based and point-based algorithms, as well as complex techniques such as joint learning strategies and hybrid contour distance loss, which poses some challenges for researchers who want to implement and apply this method.
    3. In the conclusion section, although the advantages and application scenarios of the method are summarized, the future work and improvement directions are not mentioned, which may limit readers’ further exploration and research in this field.
  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance

    YES

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html

    Formatting issues:

    1. Figure 3 in Chapter 2.3 should be centered.
    2. Tables 1 and 2 in Chapter 3.3 should be placed below the text paragraphs, not mixed with the text content.
    3. Reference 14 in the reference list has an alignment error.

    For multi-perspective review of the article: 1.Innovation: This article proposes a novel joint dense point representation segmentation architecture and introduces a new contour-aware loss function, which has some innovation.

    1. Experimental Design: The article validated the proposed method on multiple chest X-ray datasets, using comparative and ablation experiments, and provides detailed descriptions of the model implementation and training process, indicating a rigorous experimental design.
    2. Result Analysis: The experimental results show that the proposed method has significant accuracy and stability improvements on multiple datasets and can encode more discriminative and highly detailed features. The results of the comparative and ablation experiments also confirm the effectiveness of the proposed method.
    3. Article Structure: The article is well-organized, following the order of Introduction, Method, Experiment, and Conclusion, and the logical relationship between each part is clear.
    4. Shortcomings: This method requires the simultaneous use of dense-based and point-based algorithms, which may increase computational costs. Additionally, the experiments are only validated on chest X-ray datasets and need to be verified on other domains and datasets to demonstrate the method’s universality and scalability. Furthermore, some of the terminologies and methods described in the article are complex and may require further explanation.
  • Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making

    5

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

    Comprehensive analysis, including method design, experiment analysis, and result

  • Reviewer confidence

    Confident but not absolutely certain

  • [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

    The main contribution of the paper is a framework that learns both point and pixel contour representations jointly. In addition, a contour distance loss is introduced to avoid the penalization of correct predictions that fall on the contour boundaries.

  • 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. The novelty of this work lies in the joint architecture that benefits from both point and dense approaches. As a result, it acquires the segmentation precision of dense-based methods, while eliminating their topological inaccuracies. It is easily generalized to various graph segmentation networks.

    2. The experiments are well designed and the proposed approach achieves better performances than the baselines.

    3. The figures are of high quality.

  • 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.

    My major concern is that there is not enough justifications of how the parameters and hyperparameters are choose, so it is unclear to me how much the changes in the parameters and hyperparameters can affect the experiment results.

  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance

    I consider the paper is reproducible as the code is released at anonymous Github.

  • Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html

    1. I would suggest to add a justification of parameter tuning that can demonstrate the stability of the proposed framework.

    2. Some labels and legends of the figures are too small, specifically, Fig. 1, Fig. 3 and Fig. 4.

  • Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making

    6

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

    Overall, I consider this paper have solid contributions and the experiment results indicate better performance compared with the baselines.

  • Reviewer confidence

    Confident but not absolutely certain

  • [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



Primary Meta-Review

  • Please provide your assessment of this work, taking into account all reviews. Summarize the key strengths and weaknesses of the paper and justify your recommendation. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. In case of an invitation for rebuttal, clarify which points are important to address in the rebuttal.

    This work proposes a new loss function and evaluates the performance and stability of the approach well. The reviewers also appreciate the generalisability of the work and presentation of the paper. There is some concern about the increase in computational cost, which should be discussed along with the impact of the parameters on the experimental results.



Author Feedback

We’d like to thank the reviewers and meta-reviewer for their feedback. We are very grateful for the time you spent on these responses, which were both detailed and insightful. We will take your criticisms into account when editing for the camera-ready submission.



back to top