List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Yu Fang, Zhiming Cui, Lei Ma, Lanzhuju Mei, Bojun Zhang, Yue Zhao, Zhihao Jiang, Yiqiang Zhan, Yongsheng Pan, Min Zhu, Dinggang Shen
Abstract
In digital dentistry, high-quality tooth models are essential for dental diagnosis and treatment. 3D CBCT images and intra-oral scanning models are widely used in dental clinics to obtain tooth models. However, CBCT image is volumetric data often with limited resolution (about 0.3-1.0mm spacing), while intra-oral scanning model is high-resolution tooth crown surface (about 0.03mm spacing) without root information. Hence, dentists usually scan and combine these two modalities of data to build high-quality tooth models, which is time-consuming and easily affected by various patient conditions or acquisition artifacts. To address this problem, we propose a learning-based framework to generate high-quality tooth models with both fine-grained tooth crown details and root information only from CBCT images. Specifically, we first introduce a tooth segmentation network to extract individual teeth from CBCT images. Then, we utilize an implicit function network to generate tooth models at arbitrary resolution in a continuous learning space. Moreover, to capture fine-grained crown details, we further explore a curvature enhancement module in our framework. Experimental results show that our proposed framework outperforms other state-of-the-art methods quantitatively and qualitatively, demonstrating the effectiveness of our method and its potential applicability in clinical practice.
Link to paper
DOI: https://link.springer.com/chapter/10.1007/978-3-031-16443-9_22
SharedIt: https://rdcu.be/cVRyA
Link to the code repository
N/A
Link to the dataset(s)
N/A
Reviews
Review #1
- Please describe the contribution of the paper
The paper presents a method for generating high quality tooth reconstructions from CBCT scans using superresolution ideas based on implicit function networks and a curvature enhancement for the tooth crown surface based on k-NN based features and a CNN regression.
- 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.
Obtains a solution for producing quality tooth reconstructions from CBCT images without the need for the intra-oral scan and image registration to obtain an adequate accuracy.
- 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 method description is a little unclear because the descriptions of the training phase and that of how the method is used for unseen images have not been separated. Many details on the network architecture are missing, such as the number of filters in each block, the filter size, the number of training epochs, the learning rate, the optimizer used, etc. The evaluation uses a fixed test set of 20 images, which is not as convincing as a 5-fold cross-validation or multiple random splits.
- 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
Some details are missing (e.g. size and number of filters in conv layers), so the method cannot be exactly reproduced from it description.
- 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/2022/en/REVIEWER-GUIDELINES.html
In Table 2, since it is an ablation study, a more clear descrition of the ablations would be “No implicit functions, no Curvature enhacement” and so on.
- 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
6
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
Obtains state of the art results, but has some clarity issues and misses many details about training. Also, experiments could be more convincing using cross-validation.
- Number of papers in your stack
4
- What is the ranking of this paper in your review stack?
1
- 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 #2
- Please describe the contribution of the paper
The paper presents a method for 3D tooth model construction from CBCT and intra-oral images with implicit function networks (ICT). The motivation of the paper is to combine the intra-oral scans and CBCT images for better root and crowns due to limited resolution of CBCT images.
The method is validated on a dataset of 50 subjects where 20 of them are test data. The method outperforms the compared methods.
- 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.
The idea of using both modalities is very interesting. Also, implicit function networks are very popular for 3D shape reconstruction and this method is a new example of using IFN for medical data. It is applicable to several practical medical reconstruction problems.
The comparison of the method with other methods (especially HGMNet) is very useful for showing the effectiveness of using intro-oral data. I think the ablation study shows the effectiveness of the curvature enhancement and surface reconstruction.
The figures provide useful information about the method. The curvature enhancement with a separate branch is an original approach.
- 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.
Technical details of the method especially the network parameters (filter sizes, optimization parameters, etc.) are missing. The figures are very useful for presenting the method however their details are not sufficiently provided in the text. Also, more details on the experiment for the baseline models should be added.
Another major weakness is dataset size. It is very hard to gather multi-model data, however, instead of using 20train, 10 validation, and 20 test data, cross-validation would be much fairer. Also, if are there any missing teeth, dental implants, braces, etc. they should be given.
There is also reproducibility concerns about the work given in the below sections.
- 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
I have some concerns about the reproducibility of the paper. The implementation and technical details of the method are not given in the paper. In the checklist, they claim that they will publish the data, training code, and network model. However, this information is not given in the text.
- 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/2022/en/REVIEWER-GUIDELINES.html
The idea of using two modalities for 3D tooth construction and using IFN has novelty. Cross-validation would be much better for comparison with other methods and validation of the method. Although there is a section for “implementation details”, the details of the method (especially network parameters) should be given.
The running time of the method should be given.
Also, since only the crown information comes from intra-oral data, the results of the construction of the root and crown can be given also individually.
- 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
6
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The work proposes a novel method for 3D tooth model construction. The ideas of using multi-modal data and using IFN with curvature encoding look reasonable and the experimental results show their effectiveness. However, there are certain reproducibility issues, and more detailed explanations would help readers.
- Number of papers in your stack
4
- What is the ranking of this paper in your review stack?
1
- Reviewer confidence
Confident but not absolutely certain
- [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
key idea is to combine the commonly used CNN-based segmentation network with an implicit function network to generate 3D tooth models with fine-grained geometric details
- 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.
its potential applicability in clinical practice
- 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 innovation is weakness
- 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
Yes
- 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/2022/en/REVIEWER-GUIDELINES.html
- Please list the contribution or highlight of this paper
- How about the time complexity about your model compared other methods, because there is another segmentation step.
- There are lots of the reconstruct 3D shape models, why do you only select these 2 models to compare with your models? The number of SOTA is so less, which can not demonstrate this is a hot topic or Meaningful research work.
- After segmentation, each tooth is modeled by suface resconstruction, then whether it needs to be integrated into a full tooth image but not only one by one tooth model?
- 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
4
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
application and innovation
- Number of papers in your stack
5
- What is the ranking of this paper in your review stack?
3
- 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 propose a deep learning based method for reconstructing a high fidelity tooth reconstruction by combining intra-oral images and CBCT scans using the continuous parameterization of implicit function networks. The application is interesting and combining these two modalities for tooth reconstruction seems novel. The method is compared with several existing methods and ablation experiments are conducted showcasing the efficacy of the method. Weaknesses include missing technical details that affect the reproducibility of the work, missing analysis for runtime, and low-sample size. Authors are required to address the clarity concerns raised by the reviews and add missing technical details pointed out by the reviewers when preparing their camera ready.
- What is the ranking of this paper in your stack? Use a number between 1 (best paper in your stack) and n (worst paper in your stack of n papers). If this paper is among the bottom 30% of your stack, feel free to use NR (not ranked).
1
Author Feedback
N/A