List of Papers Browse by Subject Areas Author List
Abstract
Digital orthodontic workflows increasingly rely on virtual setups to predict treatment outcomes from pre-treatment intraoral scans. Although learning-based methods can automatically align teeth by estimating per-tooth rigid motions, they often neglect soft-tissue deformation, leaving post-treatment gingival morphology unknown. We propose OrthoMorphNet, a hierarchical framework for joint synthesis of post-orthodontic dentition and gingiva. The model first aligns individual teeth by fusing per-tooth geometry with global spatial context through cross-attention to regress translation and unit-quaternion rotation, supervised by a composite objective enforcing geometric regularization and physical plausibility, including collision avoidance and arch-continuity. Subsequently, the predicted dentition guides gingiva synthesis in a volumetric indicator-grid space using a Swin-transformer-based 3D encoder–decoder, followed by Marching Cubes surface extraction with a grid reconstruction loss. Experiments on a public dataset demonstrate high alignment accuracy and state-of-the-art gingiva reconstruction compared to existing volumetric baselines, supporting clinically faithful end-to-end 3D orthodontic simulation.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026/paper/1814_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{HanJi_OrthoMorphNet_MICCAI2026,
author = { Han, Ji Yong AND Shin, Woncheol AND Yang, Su AND Kim, Sujeong AND Yun, Hyunbin AND Lim, Sang-Heon AND Choi, Ho Yoon AND Ahmadi, Mobin AND Kim, Jun-Min AND Lee, Sang-Jeong AND Yi, Won-Jin},
title = { { OrthoMorphNet: Alignment-Guided Hierarchical Morphological Synthesis of Post-orthodontic Dentition and Gingiva } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 16889},
month = {September},
page = {pending}
}
Reviews
Review #1
- Please describe the contribution of the paper
This paper proposes OrthoMorphNet, a hierarchical framework that jointly predicts post-orthodontic tooth alignment and the corresponding gingival morphology from pre-treatment 3D scans, addressing a gap in prior methods that focused mainly on rigid tooth movement while ignoring soft-tissue deformation.
- 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 manuscript is well organized and clearly written. 2.The paper addresses a real clinical problem in digital orthodontics by predicting gingival soft-tissue deformation after tooth movement, rather than focusing only on rigid tooth alignment. 3.This paper provides a comprehensive evolution of the proposed method.
- 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.In Fig. 3, the visual improvements over the competing methods seem relatively limited, which makes the superiority reported in Table 2 less convincing and raises concerns about the consistency between the qualitative presentation and the numerical evaluation. 2.The approach converts intraoral scan data into volumetric representations and applies a transformer-based 3D encoder–decoder, which is technically reasonable, but this design is still relatively close to existing volumetric medical-image reconstruction pipelines. As a result, the innovation beyond integrating the two stages is not fully compelling. 3.The study relies on a public dataset, but it does not include internal or more challenging clinical cases, particularly patients with substantial gingival changes after orthodontic treatment. This limits the evidence for the model’s robustness and practical generalizability in more complex scenarios.
- 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.
- 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?
1.The manuscript is well organized and clearly written. 2.The paper addresses a real clinical problem in digital orthodontics by predicting gingival soft-tissue deformation after tooth movement, rather than focusing only on rigid tooth alignment. 3.This paper provides a comprehensive evolution of the proposed method. 4.In Fig. 3, the visual improvements over the competing methods seem relatively limited, which makes the superiority reported in Table 2 less convincing and raises concerns about the consistency between the qualitative presentation and the numerical evaluation. 5.The approach converts intraoral scan data into volumetric representations and applies a transformer-based 3D encoder–decoder, which is technically reasonable, but this design is still relatively close to existing volumetric medical-image reconstruction pipelines. As a result, the innovation beyond integrating the two stages is not fully compelling. 6.The study relies on a public dataset, but it does not include internal or more challenging clinical cases, particularly patients with substantial gingival changes after orthodontic treatment. This limits the evidence for the model’s robustness and practical generalizability in more complex scenarios.
- Reviewer confidence
Very confident (4)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Reject
- [Post rebuttal] Please justify your final decision from above.
N/A
Review #2
- Please describe the contribution of the paper
1.The paper proposes the first framework for joint synthesis of post-orthodontic dentition and gingiva. 2.The paper introduce an alignment-guided, hierarchical design that synthesizes soft tissue in an indicator-grid representation, improving noise robustness and preserving gingival topology. 3.Experiments on a public dataset demonstrate high alignment accuracy and state-of-the-art gingiva reconstruction compared to existing volumetric baselines.
- 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 paper introduces the first framework for jointly predicting post-treatment tooth alignment and the corresponding gingival morphology. Existing methods primarily focus on rigid tooth repositioning, whereas accurate gingival modeling is equally critical for realistic orthodontic simulation. In this regard, the paper pioneers a novel direction that has largely been overlooked in previous research.
- 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 paper claims to propose a hierarchical framework for the joint synthesis of post-orthodontic dentition and gingiva. However, the tooth alignment and gingiva synthesis components appear largely independent. In other words, the tooth alignment module could be readily replaced by other state-of-the-art methods without affecting the overall pipeline. 2.The tooth alignment network lacks novelty. It encodes each tooth point cloud along with the set of tooth centers to regress translation and rotation, which is a commonly adopted strategy in existing methods (e.g., TANet). Essentially, the approach mainly replaces the encoder without introducing substantial innovation. In addition, the design of the collision loss is not well justified. The paper also does not provide comparisons with state-of-the-art tooth alignment methods, and based on the reported results, the proposed method does not achieve the best performance even among the selected baselines. 3.Accurate gingival morphology is indeed essential. Existing solutions include geometry-based approaches such as Laplacian deformation. The proposed method can be viewed as a learning-based approach; however, the authors do not clearly justify the necessity of adopting a learning-based paradigm, nor do they provide comparisons with established geometry-based methods.
- 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 mention open access to source code or data, but provides a clear and detailed description of the algorithm to ensure 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.
(3) Weak Reject — marginally below the acceptance threshold, but would not mind if 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 proposes the first framework for jointly predicting post-treatment tooth alignment and the corresponding gingival morphology, representing a new research direction. However, within this framework, the tooth alignment network lacks novelty and can be easily replaced by existing state-of-the-art automatic tooth alignment methods. Consequently, the primary contribution lies in the gingiva generation network. However, this component also shows limited innovation. The method employs differentiable Poisson surface reconstruction (DPSR) as a supervisory signal and simply concatenates conditioning inputs for generation, without introducing substantial methodological advances. Furthermore, the paper does not clearly demonstrate the advantages of this deep learning–based approach for gingiva generation. To the best of my knowledge, there are already numerous geometry-based methods for gingiva modeling, as well as approaches that adapt gingival structures according to tooth movements.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Accept
- [Post rebuttal] Please justify your final decision from above.
After considering the authors’ response, I am inclined to support acceptance of this paper. While the technical contributions are relatively incremental, the work introduces a novel research direction that could stimulate further exploration and potentially benefit the community.
Review #3
- Please describe the contribution of the paper
I really like the clinical starting point of this paper in destritry. It has clear clinical relevance. Its main contribution of OrthoMorphNet is the prediction of gingival changes after orthodontic movement, introducing a new approach that models both tooth and gingival changes in tandem.
- 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 words, tables, and figures are pretty clear, readable, and well-organized in this work. -Author’s proposal and articulation of algorithmic innovation are clinically novel and detailed in this paper, and show restrained exposition of its clinical application potential. -It successfully expands orthodontic AI beyond just hard-tissue arrangement. -The authors present a highly ambitious end-to-end pipeline that effectively integrates complex 3D multimodal features of both teeth and gingiva.
- 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.
-There appear to be minor paragraph-formatting issues in the manuscript. For example, in the Introduction, the text formatting seems inconsistent after Ref. [17]. The authors are encouraged to carefully proofread the manuscript and correct these issues before publication. -A comparison with current state-of-the-art commercial software is currently missing. If feasible, including such a baseline would substantially strengthen the practical relevance of the study and help position the proposed method against existing clinical solutions. -The proposed network appears to rely primarily on visual supervision, without incorporating explicit physical boundary constraints. This raises concerns regarding clinical feasibility, as the method may produce non-physical interpenetration or unrealistic gaps between the generated gingival surface and the predicted tooth positions. -The low-resolution voxelization and subsequent reconstruction process may introduce substantial smoothing and stair-step artifacts, which can irreversibly remove high-frequency geometric details, particularly around the gingival margin. Such fine anatomical details are clinically important and should be better preserved. -Because the loss terms are fully decoupled and no backward consistency constraints are imposed, the overall framework effectively behaves as a unidirectional cascade. As a result, upstream errors in tooth arrangement may accumulate and propagate downstream, potentially amplifying inaccuracies in the final gingival prediction. -Given the extremely complex and heavy architecture (stacking multiple Transformers and 3D Convs), it is highly recommended to report the inference time and memory footprint to evaluate its clinical feasibility.
- 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.
- 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?
I recommend Accept. The paper breaks new ground by jointly predicting tooth alignment and gingival morphology, pushing orthodontic AI closer to real clinical workflows. Although the methodology has some noticeable limitations, the novelty and clinical impact of tackling soft-tissue prediction far outweigh these technical flaws. As AI in dentistry is currently in urgent need of further development, this work provides valuable new insights for researchers in the field.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Accept
- [Post rebuttal] Please justify your final decision from above.
I like their ideas. And authors successfully explain some of my questions after the rebuttal.
Author Feedback
We appreciate reviewers for their constructive comments. [Novelty & Clinical Motivation – R1,R2,R3] We appreciate the recognition of the clinical relevance of jointly predicting post-orthodontic dentition and gingival morphology. Our contribution is to formulate an alignment-guided framework where predicted post-treatment teeth explicitly condition learnable gingiva synthesis, rather than simply applying a volumetric reconstruction network. Existing orthodontic AI mainly focuses on rigid tooth repositioning, while our work addresses coupled hard/soft-tissue prediction for realistic simulation, aligner fitting, and gingival-margin/interproximal assessment. We will revise the manuscript to distinguish this contribution from conventional volumetric pipelines. [Modularity & Alignment Comparison – R2] We appreciate the concern that the alignment module could be replaced by other methods. We view this modularity as intended, since our main contribution is not a unique alignment backbone, but using predicted post-treatment dentition as a learnable deformation condition for gingiva synthesis. To ensure reproducible orthodontic planning, we use a deterministic, physically regularized alignment stage with collision and adjacency constraints, and compare with representative deterministic methods, TAligNet and TANet. Our method achieves the best ADD and rotation error, while TANet is slightly better only in translation error. The translation gap is small, 0.114mm (0.868 vs. 0.754mm), whereas our rotation improvement is larger, 3.6°, which is more relevant to angular tooth alignment and gingival-margin/interproximal morphology. [Geometry-Based Gingiva Modeling – R2] We thank the reviewer for noting geometry-based gingiva modeling and tooth-movement-driven deformation methods. We agree these are important prior works and will discuss them explicitly. Our intention is not to claim gingiva adaptation has no precedent, but to formulate it as a learnable, data-driven conditional synthesis problem: the model learns the mapping from pre-treatment gingiva/teeth and estimated post-treatment tooth positions to post-treatment gingival morphology, rather than relying on manually specified handles, boundary rules, or deformation parameters. Our current comparison focuses on reproducible 3D volumetric baselines under the same public dataset/training setting. We will state that geometry-based methods and commercial software were not included, and pursue such comparisons when feasible. [Qualitative & Quantitative Consistency – R1] We appreciate the concern that visual improvements in Fig. 3 appear limited compared with Table 2.The gains over UNETR are modest in some metrics, but consistent across MAE, R2, IoU, Dice, SDE, HD, and NC. We will revise Fig. 3 by enlarging gingival-margin/interproximal regions and using clearer distance maps. We will clarify that the gains reflect consistent local surface-error reduction rather than visually dramatic global shape changes. [Generalization & Robustness – R1,R3] We agree that validation on internal datasets and challenging cases with substantial gingival changes would strengthen robustness evidence. Our experiments use a public paired pre/post-treatment dataset to ensure reproducibility and fair comparison. We will state this limitation, avoid overclaiming clinical deployment, and pursue multi-center validation. [Physical Constraints & Efficiency – R3] The alignment stage includes collision and adjacency regularization to reduce inter-tooth interpenetration and preserve arch continuity. We agree that the gingiva stage could be improved with explicit tooth-gingiva boundary constraints and backward consistency. DPSR indicator grids were chosen for noise robustness and topology preservation, but fine details may be smoothed. We additionally profiled inference on a single NVIDIA RTX A6000 GPU: 0.214s per case with 4.27GB peak allocated VRAM (4.88GB reserved), indicating practical efficiency for preoperative planning.
Meta-Review
Meta-review #1
- Your recommendation
Invite for Rebuttal
- 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.
The paper got 1 x accept, 1 x weak accept and 1 x weak reject. I would suggest to address major reviewer’s concerns.
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation.
I would recommend to address raised issues from reviews in camera-ready version.
Meta-review #2
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation.
The reviewers showed some enthusiasm for this work being presented at the conference. Several concerns remain but the overall contributions are strong enough to justify acceptance.
Meta-review #3
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation.
After rebuttal, two reviewers suggest accepting this paper with one reviewer increasing. The other reviewer decrease the score from weak accept to reject. Based on the rebuttal as well as the recommendations of the reviewers, the paper can be accepted.
It is suggested that the authors sufficiently consider the comments raised by the reviewers and further improve the paper in the final version.