Authors

Golriz Hosseinimanesh, Farnoosh Ghadiri, Francois Guibault, Farida Cheriet, Julia Keren

Abstract

Designing a dental crown is a time-consuming and labor-intensive process. Our goal is to simplify crown design and minimize the tediousness of making manual adjustments while still ensuring the highest level of accuracy and consistency. To this end, we present a new end-to-end deep learning approach, coined Dental Mesh Completion (DMC), to generate a crown mesh conditioned on a point cloud context. The dental context includes the tooth prepared to receive a crown and its surroundings, namely the two adjacent teeth and the three closest teeth in the opposing jaw. We formulate crown generation in terms of completing this point cloud context. A feature extractor first converts the input point cloud into a set of feature vectors that represent local regions in the point cloud. The set of feature vectors is then fed into a transformer to predict a new set of feature vectors for the missing region (crown). Subsequently, a point reconstruction head, followed by a multi-layer perceptron, is used to predict a dense set of points with normals. Finally, a differentiable point-to-mesh layer serves to reconstruct the crown surface mesh. We compare our DMC method to a graph-based convolutional neural network which learns to deform a crown mesh from a generic crown shape to the target geometry. Extensive experiments on our dataset demonstrate the effectiveness of our method, which attains an average of 0.062 Chamfer Distance. The code is available at: https://github.com/Golriz-code/DMC.git Keywords: Mesh completion · Transformer · 3D shape generation.

Link to paper

DOI: https://doi.org/10.1007/978-3-031-43996-4_53

SharedIt: https://rdcu.be/dnwPW

Link to the code repository

https://github.com/Golriz-code/DMC.git

Link to the dataset(s)

N/A

Reviews

Review #1

Please describe the contribution of the paper
- Train a crown mesh generation network end-to-end
Please list the main strengths of the paper; you should write about 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.

Many of the mentioned models used in the proposed approach have been published before; however, the authors have been able to combine and integrate them for an end-to-end solution to crown mesh generation.
Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.
- Data augmentation is unclear: in particular, the 3d transformation conducted on training data are problematic, including whether they were done on individual tooth or as a whole. If on individual tooth, the augmented training data may not be a good one for training, as the ground truth crown may not fit in the context of the augmented neighboring teeth.
- Evaluation of predicted crown mesh can be improved. Besides the Chamfer distance used, the authors also need to consider whether the predicted crown will have a good contact with the opposing tooth for the optimal biting power.
- More experimental results are expected, including those shown in Fig. 5. Ablation study is also encouraged to be conducted for the work.
Please rate the clarity and organization of this paper

Good
Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance

The reproducibility of the paper is good.
Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html

Please see comments in the weakness section.
Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making

5
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

Novelty is limited and results can be further improved.
Reviewer confidence

Very confident
[Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

N/A
[Post rebuttal] Please justify your decision

N/A

Review #3

Please describe the contribution of the paper

This paper proposes a mesh completion method for dental crown design. It uses a Transformer encoder-decoder network to extract features of the input point cloud, then uses a Mesh completion layer with a differentiable Poisson solver to reconstruct the mesh from the crown points. This method shows better performance than a few baselines on a self-collected dataset.
Please list the main strengths of the paper; you should write about 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 Mesh completion layer with a differentiable Poisson solver seems to be novel.
2. The performance is good compared with baselines.
Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.

The writing contains numerous grammar errors. In its current status, it’s written in a standard that’s not ready for publication.
Please rate the clarity and organization of this paper

Poor
Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance

Seems reproducible.
Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html

For the numerous grammar errors, I suggest the authors use grammar review services, such as chatgpt to completely revise the paper.
Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making

5
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

Despite the writing issues, the technical contribution seems solid.
Reviewer confidence

Not confident
[Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

N/A
[Post rebuttal] Please justify your decision

N/A

Review #2

Please describe the contribution of the paper

A new end-to-end deep learning approach for tooth crown generation is proposed.
Please list the main strengths of the paper; you should write about 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.

A new end-to-end deep learning approach for tooth crown generation is proposed. The proposed method is sound.
Please list the main weaknesses of the paper. Please provide details, for instance, if you think a method is not novel, explain why and provide a reference to prior work.

Why the proposed method works better than the compared ones are not analyzed in detail. The current manuscript should be further polished.
Please rate the clarity and organization of this paper

Very Good
Please comment on the reproducibility of the paper. Note, that authors have filled out a reproducibility checklist upon submission. Please be aware that authors are not required to meet all criteria on the checklist - for instance, providing code and data is a plus, but not a requirement for acceptance

N/A
Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html

Why the proposed method works better than the compared ones are not analyzed in detail. Specifically, in the experiments, the analysis on the results is quite limited. Please elaborate on that.
Rate the paper on a scale of 1-8, 8 being the strongest (8-5: accept; 4-1: reject). Spreading the score helps create a distribution for decision-making

5
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

The proposed method is sound.
Reviewer confidence

Somewhat confident
[Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed

N/A
[Post rebuttal] Please justify your decision

N/A

Primary Meta-Review

Please provide your assessment of this work, taking into account all reviews. Summarize the key strengths and weaknesses of the paper and justify your recommendation. 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 proposes an end-to-end deep learning approach for tooth crown generation that outperforms several baselines on a self-collected dataset. The proposed method involves a mesh completion layer with a differentiable Poisson solver, which is considered novel. The Transformer encoder-decoder network is used to extract features of the input point cloud, and the mesh completion layer is used to reconstruct the mesh from the crown points. However, there are some issues with the data augmentation process, particularly concerning the 3D transformations conducted on training data. The authors need to clarify whether these were done on individual teeth or as a whole. Furthermore, the evaluation of the predicted crown mesh needs improvement, as the authors should consider whether the predicted crown will have good contact with the opposing tooth for optimal biting power. Additionally, an ablation study is encouraged to be conducted for the work. The manuscript should be further polished, and the authors should provide a detailed analysis of why the proposed method works better than the compared ones.

Author Feedback

Reviewer #1 Reviewer Comment: Data augmentation is unclear: in particular, the 3D transformations conducted on the training data are problematic, including whether they were done on individual teeth or as a whole. Answer: Thank you for your valuable feedback. We agree that it is important to provide clearer details about the data augmentation process. Specifically, we conducted data augmentation including 3D transformations, scaling, and rotation, on the entire dental context, which includes the master arch, opposing arch, and shell, as a unified entity. We will incorporate these updates into the revised paper to ensure a comprehensive explanation of the data augmentation process.

Reviewer Comment: Evaluation of predicted crown mesh can be improved. Answer: Regarding the evaluation of the predicted crown mesh, we appreciate your suggestion to consider the contact between the predicted crown and the opposing tooth for optimal biting power. In this work, a dental technician assesses the quality of the predicted crown. In our future research, we will address these aspects quantitatively as stated in the conclusion.

Reviewer Comment: More experimental results are expected, ablation study is also encouraged to be conducted for the work. Answer: We have indeed conducted an ablation study; however, due to space limitations, we were unable to include it in the paper. We plan to utilize the opportunity provided to expand the length of the paper by up to half a page. In the revised version, we will provide additional experimental results, including the ablation study, to further validate our approach. The ablation study consists of the following experiments:

PoinTr: This experiment employs a point completion method as the baseline.

PoinTr network with Shape as Points (SAP): Here, we utilize SAP as a separate network to reconstruct the mesh based on the point cloud generated by PoinTr.

Our proposed network Dental mesh completion (DMC) without considering the MSE (Mean Square Error) loss function.

DMC with the MSE loss function to calculate the similarity between the predicted and ground truth indicator grids. By conducting these four experiments, we provide a comprehensive evaluation of our proposed method. This evaluation covers various aspects, such as the baseline comparison, the influence of Shape as Points (SAP) as a separate network, the performance without the MSE loss function, and the effect of incorporating the MSE loss. Results from these experiments will be presented in a new Table. We will therefore revise Table 1 from the current version of the paper, to compare our approach with two distinct approaches from the literature which are PoinTr + margin line and PoinTr + graph.

Reviewers # 2, 3 Reviewer Comment: Why the proposed method works better than the compared ones? Answer: Thank you for your valuable feedback. In our method, we highlight the benefits of employing a differentiable point-to-mesh component for learning high-quality crown meshes. To ensure effective supervision, we introduce an indicator grid function, assuming access to crown meshes, and utilize the Chamfer loss on the target crown point cloud. These factors highlight the advantages of our proposed method over the compared approaches. Additionally, as mentioned earlier, we have already conducted an ablation study, and we will include its results in the revised paper. This will provide a clear understanding of why our new approach performs better. Regarding the overall polish of the paper, we will carefully review and revise the paper to ensure its clarity, readability, and coherence.

back to top

From Mesh Completion to AI Designed Crown