List of Papers Browse by Subject Areas Author List
Abstract
Medical image segmentation, particularly in multi-domain scenarios, demands precise preservation of anatomical structures across diverse representations. While deep learning has advanced this field, existing models often struggle with boundary representation, variability in organ morphology, and information loss during downsampling, limiting their accuracy and robustness. To address these challenges, we propose the Context Enhancement Network (CENet), a novel segmentation framework featuring two key innovations. First, the Dual Selective Enhancement Block (DSEB) integrated into skip connections enhances boundary details and improves the detection of smaller organs in a context-aware manner. Second, the Context Feature Attention Module (CFAM) in the decoder employs a multi-scale design to maintain spatial integrity, reduce feature redundancy, and mitigate overly enhanced representations. Extensive evaluations on both radiology and dermoscopic datasets demonstrate that CENet outperforms state-of-the-art (SOTA) methods in multi-organ segmentation and boundary detail preservation, offering a robust and accurate solution for complex medical image analysis tasks. The source code is publicly available at https://github.com/xmindflow/cenet.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2025/paper/2882_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
https://github.com/xmindflow/cenet
Link to the Dataset(s)
Synapse dataset: https://www.synapse.org/#!Synapse:syn3193805/wiki/89480
ACDC dataset: https://www.creatis.insa-lyon.fr/Challenge/acdc/databases.html
PH2 dataset: https://www.fc.up.pt/addi/ph2%20database.html
HAM1000 dataset: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T
BibTex
@InProceedings{BozAfs_CENet_MICCAI2025,
author = { Bozorgpour, Afshin AND Kolahi, Sina Ghorbani AND Azad, Reza AND Hacihaliloglu, Ilker AND Merhof, Dorit},
title = { { CENet: Context Enhancement Network for Medical Image Segmentation } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
year = {2025},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15960},
month = {September},
page = {120 -- 129}
}
Reviews
Review #1
- Please describe the contribution of the paper
This paper proposes a novel medical image segmentation framework named CENet, aiming to improve the ability of boundary details and semantic representation in multi-scale and multi-organ segmentation tasks. The main contributions of the paper include two innovative points: Firstly, a DSEB is proposed. By enhancing the boundary details in skip connections, it improves the detection ability of small organs. Secondly, a CFAM is designed, which adopts a multi-scale design to maintain spatial integrity, reduce feature redundancy, and avoid overly enhanced representations. Experiments show that CENet outperforms existing state-of-the-art methods on multiple radiology and skin lesion datasets, providing a more accurate and robust solution for multi-organ segmentation and boundary detail preservation.
- 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.
The HD95 of CENet in the organ segmentation task has been significantly improved, far surpassing other methods.
- 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 The writing of this manuscript is rather redundant. For example, in the Introduction section, a large amount of space is used to elaborate on the background, which is redundant.
2 When the author points out the existing problems, the cited references are relatively outdated. For instance, [14, 17] are from 2023, and [27] is from 2022.
3 In Table 3, the improvements brought about by the introduction of different modules are not significant. It is recommended that the author conduct significance statistics and present the mean and variance of the performance of each row in the ablation experiment to rule out the influence of random fluctuations.
4 The author has set different training configurations for different datasets. Are these the best results after combined experiments? Are the other methods in the comparative experiments also based on the same settings?
5 In Table 1, why does CENet perform relatively poorly on the Gallbladder (Gal.) and Stomach (Sto.)? The author claims in the abstract that “(DSEB) integrated into skip connections enhances boundary details and improves the detection of smaller organs in a context-aware manner.” However, for the Gallbladder (Gal.), it is a relatively small organ, but the method in this paper shows poor performance on it.
- Please rate the clarity and organization of this paper
Poor
- 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.
(3) Weak Reject — could be rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
see weakness
- 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
-
Introduced to enrich skip connections with stronger boundary-focus and small-organ detection capabilities. By combining the Feature Edge Amplifier (FEA) and Differential Attention (DiffAtt), DSEB helps preserve fine-grained spatial details, especially around edges, while highlighting salient context and suppressing irrelevant regions.
-
Proposed in the decoder to refine multi-scale feature representation. Through its Channel Calibration Unit (CCU), Multi-scale Contextual Aggregator (MCA), weighted Non-local Block (wNLB), and enhanced MLP, CFAM reduces over-enhancement and redundancy while capturing both local and global contextual information.
-
By integrating DSEB into skip connections and employing CFAM in the decoder, the network significantly mitigates information loss typically seen in downsampling. CENet thereby improves boundary delineation, handles morphological variations of organs, and enhances the clarity of small or intricate anatomical structures.
-
Extensive experiments on radiology (Synapse, ACDC) and dermoscopic (PH2, HAM10000) datasets show that CENet not only outperforms various CNN- and Transformer-based segmentation methods but also preserves boundary information more effectively. This results in heightened accuracy for multi-organ and lesion segmentation, demonstrating its robustness and generalizability.
-
- 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.
-
The Dual Selective Enhancement Block (DSEB) combines edge amplification and differential attention to better highlight boundaries and preserve fine-grained details, thereby improving segmentation performance for small and irregular anatomical structures.
-
The Context Feature Attention Module (CFAM) enhances contextual understanding via its channel calibration, multi-scale aggregation, non-local denoising, and MLP-based spatial recalibration. This multi-level processing helps maintain spatial integrity and reduces feature redundancy.
-
Extensive experiments on both radiology and dermoscopic datasets demonstrate that CENet outperforms state-of-the-art methods in terms of accuracy and boundary detail preservation, indicating its effectiveness across diverse medical imaging tasks.
-
By leveraging a Pyramid Vision Transformer V2 (PvT-V2) backbone and introducing specialized modules (DSEB, CFAM) in the encoder-decoder pipeline, CENet attains a good equilibrium between global receptive fields and targeted local boundary refinement.
-
In terms of narrative flow, the paper is well-structured and logically coherent, guiding readers smoothly from the research motivation to the method description, experimental validations, and final conclusions. Moreover, each module’s role and innovation are clearly articulated, enabling readers to grasp how the Dual Selective Enhancement Block and Context Feature Attention Module contribute to the overall effectiveness of the proposed framework.
-
- 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.
-
The paper reports parameter and FLOP counts but does not extensively discuss inference speed, GPU memory requirements, or potential deployment constraints. This information could be crucial for practitioners aiming to adopt the method in resource-limited or real-time clinical settings.
-
While each building block (DSEB, CFAM) is described thoroughly, the transitions between them (e.g., how precisely the skip connections integrate with the CFAM in complex scenarios) could benefit from further clarification. A more explicit delineation might help readers replicate or adapt the design more easily.
-
- 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?
Refer to the Strengths and Weaknesses section.
- 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 #3
- Please describe the contribution of the paper
This paper proposes CENet (Context Enhancement Network), a method aimed at improving context-aware medical image segmentation. It incorporates the DSEB module to enhance detailed features and reduce irrelevant context. And the proposed CFAM module is used to refine the mult-iscale features during up-sampling process. The authors conduct extensive experiments on four datasets to evaluate the effectiveness of the proposed method.
- 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.
- This paper is generally well-written and understandable.
- The proposed method ahieves strong SOTA peformance on four datasets.
- The proposed DSEB module can enhance boundary details in segmentation tasks, and can also be easily integrated into the skip connections of other existing segmentation models.
- 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.
- The proposed CFAM module is extremely complicated as it contatins so many components. Specifically, a single decoder includes MCA and SRM, while the MCA further contains wNLB, CCU, and various split and concatenation operations. It is difficult to know which component is actually contributing to the performance improvements.
- In the ablation study, the analysis of the Spatial Recalibration Module (SRM) seems to be missing. Moreover, since the MCA module is such complex, it need more detailed analysis. For example, within the MCA module, the necessity of splitting the input F’ into four parts followed by concatenation should be validated through ablation studies.
- As for the “C-MASP “ in the SRM module shown in Fig. 1, it seems that the paper does not provide a clear definition of it.
- 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?
The proposed method achieves impressive SOTA performance on various medical segmentation datasets. It will be more convincing if the authors provide a more detailed analysis about the proposed complex module.
- 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
Author Feedback
N/A
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”.
The reviewers appreciated the strong technical contributions of your work, particularly the integration of the DSEB and CFAM modules, and the clear improvements demonstrated across multiple medical imaging datasets. Your paper was recognized for its methodological novelty, thorough experimentation, and generally strong writing and organization.
Some areas for further strengthening include clarifying the complexity and individual contributions of CFAM’s components, providing more detailed ablation analyses, and addressing the observed inconsistencies in specific organ segmentation performances. While these points do not undermine the overall contribution, we encourage you to address them carefully during your final revision.