List of Papers Browse by Subject Areas Author List
Abstract
Cardiac motion over a cardiac cycle is crucial for quantifying regional function and is strongly affected by cardiovascular diseases. Since temporally dense cardiac mesh sequences are difficult to obtain in practice, we focus on leveraging the more accessible end-diastolic frame to infer a full-cycle sequence. Due to strong regional and disease-specific differences, traditional methods often oversmooth the data by relying on generative models that are optimized for global patterns. To address this problem, we propose Region-Aware and Phenotype-Adaptive Bi-Ventricular Cardiac Motion Synthesis (RePCM) for single frame Bi-ventricular mesh motion completion. In Stage I, a reconstruction network learns vertex-wise motion descriptors and clustering yields a data-driven functional partition, providing an explicit motion derived region structure. In Stage II, a Region-Specific Injection Module enforces masked, synchronized region exchange within a conditional VAE, preserving localized specific dynamics and restricting cross-region mixing. A Phenotype-Adaptive Mixture-of-Experts prior conditioned on ED shape uses anatomy-guided cues to model latent motion trends and capture inter-disease variability. Experiments on three datasets covering different cardiovascular diseases show consistent gains in geometric and functional metrics and improved preservation of region specific dynamics. Our code is available at https://github.com/yangxuan-nus/RePCM.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026/paper/2090_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
https://github.com/yangxuan-nus/RePCM
Link to the Dataset(s)
N/A
BibTex
@InProceedings{YanXua_RePCM_MICCAI2026,
author = { Yang, Xuan AND Yuan, Xiaohan AND Li, Hao AND Chen, Lingyu AND Liu, Yanan AND Li, Lei},
title = { { RePCM: Region-Specific and Phenotype-Adaptive Bi-ventricular Cardiac Motion Synthesis } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 16890},
month = {September},
page = {pending}
}
Reviews
Review #1
- Please describe the contribution of the paper
The authors proposed a region-aware conditional generative framework for completing biventricular mesh motion, where motion descriptors are clustered into shared regions and structured via an adjacency prior. By integrating these priors into a conditional VAE with a phenotype-adaptive mixture-of-experts, the model has more temporally consistent dynamics.
- 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.
Firstly, the authors separated region discovery via clustering from generation, which improves interpretability. Secondly, the method can work with multiple pathologies. Moreover, the authors compared metrics for the proposed method with previous approaches and validated their method on public datasets, which is a positive point.
- 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 method itself can be described more clearly. For example, the authors used existing methods such as DyMeshVAE or FiLM, but they could briefly explain the key ideas behind these techniques for completeness. Additionally, computation time is not reported or compared with other methods, which would be important for assessing practical applicability. The authors should also more clearly state the limitations of their approach in the conclusion, such that the use of a binary adjacency matrix may oversimplify relationships between regions. Finally, some minor grammar issues should be corrected. The authors are encouraged to publicly share their code to improve reproducibility.
- 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.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- 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/2026/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?
This is a novel methodological work that outperformed previous methods, but some clarity can be added, including limitations.
- 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
Authors propose a region-specific and phenotype-adaptive motion synthesis framework relying on VAEs and a mixture of experts
- 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.
Authors build on previous work to further improve motion synthesis results, using multiple public datasets for training/validation
- 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 utility of this work in clinical practice is questionable. While limited/inconsistent temporal sampling is a problem, even limited temporal sampling would be a great improvement over relying only on ED, where extrapolating motion to the full cardiac cycle is extremely challenging. But as a methodological exercise, this is a valid and strong work
- Standard deviations are reported, but cross-validation is not performed.
- 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 provide sufficient information for reproducibility.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- 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/2026/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?
- Strong contribution and improved results in an evolving field.
- 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
The paper proposes a Region-Aware and Phenotype-Adaptive Bi-Ventricular Cardiac Motion Synthesis (RePCM) framework. The key contribution lies in decomposing cardiac motion into data-driven functional regions and explicitly modeling region-level interactions through a region-constrained attention mechanism. Additionally, the method introduces a shape-conditioned, phenotype-adaptive Mixture-of-Experts (MoE) prior to capture disease-specific motion variability. Together, these designs enable the generation of anatomically consistent and temporally coherent 3D+t cardiac motion sequences with improved preservation of region-specific dynamics across different cardiovascular conditions. The experimental evaluation is relatively comprehensive, including multiple datasets, comparisons with several baselines, and ablation studies to validate key components of the proposed framework.
- 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.Accurate cardiac motion estimation is a clinically important problem. The proposed research is highly relevant to the medical imaging community. The authors clearly motivate the need for region-aware and disease-aware motion modeling. 2.The proposed framework is novel in combining data-driven functional partitioning, region-constrained interaction, and a phenotype-adaptive MoE prior, which provides a structured way to model spatially heterogeneous and disease-specific cardiac motion. 3.The method explicitly models region-level dynamics, which is more aligned with physiological behavior compared to global latent modeling and helps preserve localized motion patterns. 4.The experimental evaluation is relatively comprehensive. It conducted on large-scale, multi-center datasets, includes multiple baselines and quantitative metrics, provides ablation studies to validate key components (region partition and MoE design) 5.The model demonstrates consistent improvements across metrics and shows better temporal consistency and anatomical plausibility. 6.The inclusion of phenotype-aware analysis (e.g., expert usage and disease-specific dynamics) adds interpretability and provides useful insights into how the model captures inter-disease variability.
- 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 method pipeline is relatively complex, involving multiple components (region partitioning, masked attention, MoE prior, and anchor-based decoding). However, the paper does not mention plans to release code, which raises concerns about reproducibility and practical adoption.
2.The design lacks clear justification for some components. For example, the interaction between modules (e.g., region partitioning and MoE prior) is not fully analyzed, and it is unclear how much each contributes to the final performance.
3.The performance (in Table 1) improvements over prior methods, although consistent, are moderate.
- 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 provide sufficient information for reproducibility.
- Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?
N/A
- 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/2026/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 — marginally above the acceptance threshold, but would not mind if rejected, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The paper is novel and supported by relatively comprehensive experiments. However, the performance gains are moderate, and the lack of code release raises concerns about reproducibility. If the authors can provide the code, the work would be significantly strengthened and more suitable for acceptance.
- 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
We gratefully thank the reviewers and AC for their remarks and suggestions. We appreciate the encouraging comments, such as “novel methodological work” and “region discovery via clustering improves interpretability” from R1, “valid and strong work” and “strong contribution and improved results” from R2, and “novel in combining data-driven functional partitioning, region-constrained interaction, and a phenotype-adaptive MoE prior” with a “relatively comprehensive” evaluation from R3.Our responses to the questions and concerns are given below:
(1) Q: Reproducibility and code release. (R1/R2/R3) A: We will release the code and model weights after publication to improve reproducibility and promote further research on cardiac motion generation.
(2) Q: Clinical positioning of ED-only completion. (R2) A: ED-only completion serves as a challenging sparse-conditioning setting for motion completion. It is relevant when dense 3D+t motion is unavailable or incomplete, such as in cases with missing frames, partial annotations, virtual cohort generation, or simulation initialization. Since no ES or intermediate-frame information is provided, this setting also acts as a strict lower-bound benchmark to test whether the proposed method can infer disease-related motion patterns from anatomy alone. Clinically richer settings, such as sparse multi-frame conditioning, will be discussed as future extensions.
(3) Q: Clarification of adopted components. (R1/R3) A: We will better explain the roles of the adopted components in the final version. DyMeshVAE follows the idea of shape-motion disentanglement and learns motion-aware vertex descriptors through reconstruction, which are then used for functional region discovery. FiLM uses conditioning information to modulate intermediate features. In our framework, it injects region context into anchor features for region-conditioned modulation.
(4) Q: Contribution of each module and their interaction. (R3) A: We already include ablations for the main modules and will make their interpretation clearer. The R=16/24/32 experiments are conducted without MoE, which isolates the effect of region-aware design and region granularity from the MoE prior. The E=4/6/8 experiments further analyze the phenotype-adaptive prior on top of the region-aware design. These two components address complementary aspects of motion synthesis: the region module constrains local inter-region motion exchange and reduces over-smoothing across functionally different regions, while the MoE prior models disease/phenotype-related motion distributions. We will explicitly clarify this relationship in the final version.
(5) Q: Computational cost and practical applicability. (R1/R3) A: Under the unified implementation, RePCM has lower computational cost than the adapted baselines because the major interactions are performed in compressed token spaces rather than dense vertex space. Specifically, RePCM represents the mesh with K=512 anchor tokens and performs region exchange only over R=16 pooled region tokens. This design avoids expensive dense vertex-level interactions while preserving explicit localized motion modeling. We will report the profiling results in the final version.
(6) Q: Validation protocol. (R2) A: Experiments use a patient-level split to avoid subject leakage. For each test subject, 15 generated samples are averaged to evaluate robustness to stochastic generation. We agree that full cross-validation can further assess split sensitivity and will consider it for further validation.
(7) Q: Limitations and details. (R1) A: We will discuss the simplification of continuous mechanical coupling by the binary adjacency matrix and explore soft or learnable adjacency as future work. We will also correct grammar, figure labels, and formatting issues in the final version.
Meta-Review
Meta-review #1
- Your recommendation
Provisional Accept
- Please justify your decision. 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 paper is a strong methodological contribution. The core idea of structuring motion generation around data-driven functional regions and disease-specific priors is novel and sound. The evaluation is strong and supports the claims. While complexity and reproducibility are minor concerns, the conceptual novelty is sufficient for acceptance.