List of Papers Browse by Subject Areas Author List
Abstract
Accurate segmentation of Langerhans cells (LCs) in corneal confocal microscopy (CCM) images is crucial for diagnosing and monitoring various ocular and systemic diseases. However, existing segmentation methods often struggle with the misidentification of activated LCs and inaccurate boundary delineation due to their complex morphological features and background noise. In this paper, we propose a novel segmentation framework, MorphoBoost, which integrates morphology-driven data augmentation and boundary optimization loss to address these challenges. MorphoBoost employs a “localization before segmentation” strategy, enhancing the diversity of activated LCs via spatial and appearance transformations, and refining segmentation boundaries through pixel-level and image-level optimizations. Our methods achieve state-of-the-art performance in segmenting both LCs types, especially activated ones. It establishes a new benchmark with a 17.10\% increase in the Dice coefficient and a 5.71 decrease in modified Hausdorff distance over previous methods. This is bolstered by validation on clinical data.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2025/paper/1356_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
N/A
Link to the Dataset(s)
N/A
BibTex
@InProceedings{LiHon_MorphoBoost_MICCAI2025,
author = { Li, Hongshuo and Dong, Ankai and Zhang, Tiande and Zhou, Shijia and Zheng, Yalin and Mou, Lei and Zhao, Yitian},
title = { { MorphoBoost: Morphology-Driven Boundary Enhancement Model for Accurate Segmentation of Langerhans Cells in Corneal Confocal Microscopy Images } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
year = {2025},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15972},
month = {September},
page = {381 -- 391}
}
Reviews
Review #1
- Please describe the contribution of the paper
The paper introduces MorphoBoost, a framework for segmenting Langerhans cells (LCs) in Corneal Confocal Microscopy (CCM) images. It addresses the challenge of segmenting morphologically complex activated LCs and refining boundaries through three key components: a Morphology-Driven Data Augmentation (MD-Aug) module that increases training diversity using spatial transformations like rotation and Moving Least Squares warping along with appearance changes; a Dilation-Guided Localization (DG-Loc) module that implements a “localization before segmentation” strategy by generating region-of-interest images from ensemble predictions; and a Boundary Optimization Loss (BO-Loss) that combines pixel-level and image-level terms to improve segmentation boundary accuracy.
- 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 main strengths of the paper lie in its methodological improvements, particularly in data augmentation and the segmentation pipeline, which together achieve state-of-the-art performance. This success is largely due to the framework’s focus on challenging cases of activated Langerhans cells. The effectiveness of each component is supported by a detailed ablation study, and the method has been thoroughly evaluated against existing approaches.
- 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 lacks sufficient details about the dataset used, and since the dataset is not publicly available, this limits the reproducibility and impact of the contribution. Additionally, some of the baseline methods, such as TransUNet and MedNeXt, show unexpectedly low performance both visually and quantitatively, raising concerns about the implementation of the compared methods. The proposed pipeline also relies on multiple cascaded components, and if a segmentation network is applied after ROI generation, it would be appropriate to compare alternative segmentation methods at the same stage of the pipeline. There are reproducibility concerns since neither the code, nor the dataset are being made public.
- 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 submission does not provide sufficient information for reproducibility.
- 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 tackles a relevant clinical problem with a well-motivated and technically robust approach. MorghoBoost integrates multiple components to effectively adress activated LC segmentation and boundary refinement, achieving state-of-the-art results. There are reproducibility concerns since neither the code, nor the dataset are being made public. The model specifics are not disclosed. Despite concerns about the validation of the method and dataset transparency, this contribution is suited for MICCAI.
- Reviewer confidence
Somewhat confident (2)
- [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
The method introduces a morphology-driven data augmentation module and a boundary optimization loss function, embedded within a “localization before segmentation” strategy. The augmentation simulates realistic morphological variations using spatial and appearance transformations, while the loss integrates pixel-level and image-level components to refine segmentation boundaries. Extensive experiments on both public and clinical datasets demonstrate significant improvements over state-of-the-art methods, especially for the more challenging activated LCs.
- 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.
Novel data augmentation strategy: The proposed morphology-driven augmentation simulates diverse morphological changes using spatial and appearance transformations guided by biologically relevant structures, which enhances model generalization for rare or challenging cell types. Boundary optimization loss: The paper introduces a carefully designed loss that operates both at the pixel and image level to improve boundary accuracy, addressing a common challenge in CCM image segmentation. Effective use of “localization before segmentation” paradigm: This design choice, combined with the DG-Loc module, enables focused feature extraction in regions of interest, improving performance and efficiency. Strong experimental results: The proposed method achieves substantial improvements compared to multiple baselines on both non-activated and activated LCs.
- 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.
Limited generalization validation: While two datasets (one public and one clinical) are used, both are from the same modality and anatomical region. The method’s applicability to other cell types or imaging modalities remains unclear. Ablation study could be expanded: Although the contributions of the MD-Aug and BO-Loss modules are mentioned, a more detailed ablation with quantitative results for each component would better support the claimed contributions. While the proposed method incorporates novel components (e.g., morphology-driven augmentation and boundary optimization loss), the paper lacks a clear and explicit description of the specific problem these innovations are designed to solve. In particular, the introduction does not sufficiently articulate the underlying challenges of Langerhans cell segmentation—such as why existing methods fail in identifying activated cells or delineating boundaries—and how these challenges motivate the design of each module. Strengthening the problem formulation and clearly linking it to the proposed solutions would significantly improve the clarity and impact of the work.
- 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.
- 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.
(5) Accept — should be accepted, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The methodology is original, well-structured, and clearly presented. The paper addresses a real clinical need and demonstrates a thoughtful integration of prior biological knowledge into the modeling process. While the work would benefit from broader validation beyond a single modality and a more detailed ablation study, the current results and contributions are sufficient to support acceptance. Overall, the paper is strong in both technical merit and practical relevance, and I recommend acceptance.
- 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
The paper proposes a novel MorphoBoost framework for Langerhans cell segmentation in corneal confocal microscopy images. The proposed framework includes a morphology-driven data augmentation module (MD-Aug), a RoI image generation module, and a loss function for boundary optimization.
- 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 paper provided a detailed description of the proposed framework with its constructing modules along with their functions in the framework. The paper also provided a comprehensive literature review and pointed out the challenge of the current SOTA.
- 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.
While the paper demonstrated the superior performance of the proposed framework over the other SOTA, the reason for these improvements has not been clearly explained in the discussion section of the paper. In addition, the drawback or limitations of the proposed approach should have been discussed in the paper as well.
- 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.
- 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.
(5) Accept — should be accepted, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The novelty of this work is clearly demonstrated in the paper with a notable improvement in the performance compared to the other SOTA. In addition, enough detailed description is provided for reproducibility
- 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
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”.
All three reviewers recommend accepting this paper, considering its clear writing, good technical contributions, and potential impact on related research areas. Therefore, a Provisional Accept decision is given.