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

Abstract

Medical images are usually collected from multiple domains, leading to domain shifts that impair the performance of medical image segmentation models. Domain Generalization (DG) aims to address this issue by training a robust model with strong generalizability. Re- cently, numerous domain randomization-based DG methods have been proposed. However, these methods suffer from the following limitations: 1) constrained efficiency of domain randomization due to their exclusive dependence on image style perturbation, and 2) neglect of the adverse effects of over-augmented images on model training. To address these is- sues, we propose a novel domain randomization-based DG method, called content style augmentation (ConStyX), for generalizable medical image segmentation. Specifically, ConStyX 1) augments the content and style of training data, allowing the augmented training data to better cover a wider range of data domains, and 2) leverages well-augmented features while mitigating the negative effects of over-augmented features during model training. Extensive experiments across multiple domains demonstrate that our ConStyX achieves superior generalization performance. The code is available at https://github.com/jwxsp1/ConStyX .

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2025/paper/0279_paper.pdf

SharedIt Link: Not yet available

SpringerLink (DOI): Not yet available

Supplementary Material: Not Submitted

Link to the Code Repository

https://github.com/jwxsp1/ConStyX

Link to the Dataset(s)

https://zenodo.org/records/8009107

BibTex

@InProceedings{CheXi_ConStyX_MICCAI2025,
        author = { Chen, Xi AND Shen, Zhiqiang AND Cao, Peng AND Yang, Jinzhu AND Zaiane, Osmar R.},
        title = { { ConStyX: Content Style Augmentation for Generalizable Medical Image Segmentation } },
        booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
        year = {2025},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 15964},
        month = {September},
        page = {100 -- 110}
}

Reviews

Review #1

  • Please describe the contribution of the paper

    The paper proposes a novel content-style augmentation method (ConStyX) that significantly improves the generalization capability of medical image segmentation models by combining both content and style augmentation. The idea of moving feature vectors in the feature space provides a fresh perspective for DG in medical image segmentation.

  • Please list the major strengths of the paper: you should highlight 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 is clearly described, particularly the design of the Deep Feature Augmentation (DFA) and Augmented Feature Utilization (AFU) modules, which are well-justified and experimentally validated.
    2. The experiments are comprehensive. Ablation studies effectively demonstrate the contributions of each module.
    3. The paper is well-structured.
  • Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
    1. For DFA, more details on how directions (e.g., gradient directions) are selected would help readers better understand the method.
    2. For AFU, is there a theoretical or empirical basis for the threshold selection (e.g., τ=0.6)?
    3. A discussion on computational cost, existing limitations and and potential improvements of your method is needed.
  • Please rate the clarity and organization of this paper

    Satisfactory

  • 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 authors claimed to release the source code and/or dataset upon acceptance of the submission.

  • Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2025/en/REVIEWER-GUIDELINES.html

    N/A

  • 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.

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

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

    The paper introduces a novel content-style augmentation method that significantly enhances the generalization of medical image segmentation through feature-space manipulation and weighting strategies.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.

    N/A

  • [Post rebuttal] Please justify your final decision from above.

    N/A



Review #2

  • Please describe the contribution of the paper

    This paper proposes ​ConStyX, a domain randomization-based method for generalizable medical image segmentation. This paper augments ​both content and style by perturbing deep features along two directions (intra-class variation and feature gradient guidance). Also this paper dynamically re-weights augmented features during training using cosine similarity between original and augmented features and prediction confidence of the model on augmented features.This paper achieves SOTA performance on fundus datasets.

  • Please list the major strengths of the paper: you should highlight 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) Novel data augmentation method. Augmenting both content and style in feature space is interesting. (2) SOTA performance. This paper achieves SOTA performance. (3) Controlled perturbation mechanism. Combining statistical distributions (covariance-based) and gradient guidance ensures​semantic consistency while diversifying features.

  • Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.

    (1) Limited explanation. The assumption that “feature movement preserves semantics” lacks theoretical justification. (2) Limited dataset. This paper only conduct experiments on fundus dataset. Generalizability to other medical modalities, such as CT or MRI, remains unproven. (3) Limited visualization. This paper does not provide any data augmented images. These images need to be provided to prove the author’s point.

  • 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 authors claimed to release the source code and/or dataset upon acceptance of the submission.

  • Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2025/en/REVIEWER-GUIDELINES.html

    N/A

  • 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.

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

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

    This paper need to provide more explanation about the assumption in the paper and more experiment on more datasets. Also data-augmented images is needed to show the point of the author.

  • Reviewer confidence

    Very confident (4)

  • [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.

    N/A

  • [Post rebuttal] Please justify your final decision from above.

    N/A



Review #3

  • Please describe the contribution of the paper

    With adjusting the deep features in the right directions by appropriate extents, ConStyX enhances the training data’s content and style to ensure that the source domain data that can cover a broader array of unseen domains. A strategy for utilizing augmented features has been developed to measure how much these features contribute to model training.

  • Please list the major strengths of the paper: you should highlight 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.

    An effective and controllable content-style augmentation method for disc and cup segmentation.

  • Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.

    Experiments Lack generality and broader applicability

  • 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 authors claimed to release the source code and/or dataset upon acceptance of the submission.

  • Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2025/en/REVIEWER-GUIDELINES.html

    (1)The experimental materials requires additional enhancement.It is more conducive to understanding and reproducing the model if the paper could provide adequate and comprehensive details about the experiments. For example,Within the experimental configuration, which one was assigned as the source domain?

    (2)References to preprints should be replaced with their officially published editions(if available).

    (3)‘Through these movements, deep features are appropriately augmented along the correct directions with proper degrees’. How to assess the suitability of direction and degrees?

    (4)‘Style and Content’ Are they entirely distinct from one another? For example,’different optic discs may contain distinct optic vessels in fundus images.’Why it can not be seen as a style?

  • 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.

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

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

    (1)Insufficient literature review: the lack of discussion with the most recent methodologies. DOI:10.1109/TMI.2022.3224067(2023 CSDG) DOI:10.1016/j.bspc.2024.106801(2024 PCSDG) Although the data sets show some variations, the two literature all address the domain shift problem in SSDG for Medical Image Segmentation,full discussion which would enrich the narrative, it would provide readers with a broader perspective on data augmentation and SSDG (2)Experiments lack generality and broader applicability the experimental validation is limited to a narrow set of tasks. The evaluation is primarily focused on optic cup and optic disc segmentation. However, broader claims of generalization across unseen domains remain unsubstantiated.leaving uncertainty regarding its applicability to more diverse modalities such as CT scans, brain MRIs, or histopathological images.

    Comparison with baselines is incomplete: How does it compare with other disentanglement methods beyond CCSDG?

  • Reviewer confidence

    Very confident (4)

  • [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.

    N/A

  • [Post rebuttal] Please justify your final decision from above.

    N/A



Author Feedback

We sincerely thank all the reviewers and AC for their constructive comments and suggestions. We will revise our manuscript accordingly and release our code on GitHub.



Meta-Review

Meta-review #1

  • Your recommendation

    Provisional Accept

  • If your recommendation is “Provisional Reject”, then summarize the factors that went into this decision. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. You do not need to provide a justification for a recommendation of “Provisional Accept” or “Invite for Rebuttal”.

    N/A



back to top