List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Zhengwang Wu, Jiale Cheng, Fenqiang Zhao, Ya Wang, Yue Sun, Dajiang Zhu, Tianming Liu, Valerie Jewells, Weili Lin, Li Wang, Gang Li
Abstract
The cerebellum (i.e., little brain) plays an important role in motion and balances control abilities, despite its much smaller size and deeper sulci compared to the cerebrum. Previous cerebellum studies mainly relied on and focused on conventional volumetric analysis, which ignores the extremely deep and highly convoluted nature of the cerebellar cortex. To better reveal localized functional and structural changes, we propose cortical surface-based analysis of the cerebellar cortex. Specifically, we first reconstruct the cerebellar cortical surfaces to represent and characterize the highly folded cerebellar cortex in a geometrically accurate and topologically correct manner. Then, we propose a novel method to automatically parcellate the cerebellar cortical surface into anatomically meaningful regions by a weakly supervised graph convolutional neural network. Instead of relying on registration or requiring mapping the cerebellar surface to a sphere, which are either inaccurate or have large geometric distortions due to the deep cerebellar sulci, our learning-based model directly deals with the original cerebellar cortical surface by decomposing this challenging task into two steps. First, we learn the effective representation of the cerebellar cortical surface patches with a contrastive self-learning framework. Then, we map the learned representations to parcellation labels. We have validated our method using data from the Baby Connectome Project and the experimental results demonstrate its superior effectiveness and accuracy, compared to existing methods.
Link to paper
DOI: https://doi.org/10.1007/978-3-031-43993-3_42
SharedIt: https://rdcu.be/dnwNN
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 proposes a method for cortical surface-based analysis of the cerebellar cortex. First the cerebellar cortex is reconstructed from segmentation data. Next, a weakly supervised graph CNN is used to parcellate the cerebellar cortical surface into anatomically meaningful regions. Contrastive learning is used to learn a representation of the cerebellar cortical surface that is then mapped to parcelation labels using a MLP. Experiments on data from the Baby Connectome Project show good performance.
- 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 method is interesting and well justified, providing a better formulation than the atlas-based methods or methods that map the surface to a sphere
- An accurate cerebelum reconstruction method is proposed
- The parcelation method is well designed ; I like the idea and the use of GCN and contrastive learning to acquire the latent space representation that is then used for classification
- 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.
- Post-processing : to ensure spatial consistency of patches, a graph-cut method is used as post-processing; Would there be a way to incorporate spatial consistency into the MLP formulation using some regularization similar to what the graph-cut is using
- Validation is a bit weak - 10 subjects from BCP dataset are used.
- 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
Section 3.1 provides some details on the network architecture. Details on the hyperparams used in training are not provided. Some details are missing Are the authors planning to make the code
- 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
I found the paper well written. One thing to improve would be to try to unify the MLP and graph cut approach into one formulation.
- 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 paper presents a novel method for cerebelum parcelation. The approach is well designed and works well. I would have liked to see the methods more integrated (at least the last two steps on classification and postprocessing with GC) . The validation is not very strong due to the limited number of images.
- 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 #2
- Please describe the contribution of the paper
The paper proposes a cortical surface-based analysis of the cerebellar cortex, which involves reconstructing the cerebellar cortical surfaces and automatically parcellating them into anatomically meaningful regions using a weakly supervised graph convolutional neural network. Unlike previous studies, the proposed method deals directly with the original cerebellar cortical surface, without requiring registration or mapping to a sphere. The authors validated their approach using data from the Baby Connectome Project and achieved superior effectiveness and accuracy compared to existing 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 paper is well-written and clear, presenting a novel approach to analyzing the cerebellar cortex that addresses the limitations of previous volumetric analysis methods. The proposed pipeline is a combination of different pre-existing blocks, but the approach is still interesting and effective. The method involves reconstructing the cerebellar cortical surfaces and using a weakly supervised graph convolutional neural network to automatically parcellate them into anatomically meaningful regions. The authors validated their approach using data from the Baby Connectome Project, and the results demonstrate its superior effectiveness and accuracy compared to existing methods.
- 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 validation of the approach is quite limited. Although the authors used data from the Baby Connectome Project to validate their method, more subject could have be used to strengthen the conclusions drawn in this paper. Additionally, some figures are not clear and could be improved to better demonstrate the proposed method. Finally, while the general methodology is described well, many details on the actual implementation of the proposed method are not provided, which could limit reproducibility and the ability of others to use and build on this 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
Several crucial details, including hyperparameters, network architecture, and learning rates, are missing, and the code is not available, which significantly undermines the reproducibility of this work.
- 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
My main concern is related to the experimental setting. I found the validation and experimental settings to be very limited, as only a small number of subject were used, and the comparison with state-of-the-art methods was limited to vary basic approaches.
I also have a concern about the initial step of the pipeline that you have proposed in Fig. 1. Can you provide more information on why you have chosen this specific pipeline instead of a standard one to reconstruct the inner cerebellar surface? Additionally, have you considered the disadvantages/advantages of the block you are proposing compared to existing methods such as iBEAT V2.0?
Furthermore, I found the pipeline described in Figure 3 to be unclear. For instance, the ResNet block is actually a graph CNN right? But this is not clear from the figure. Thus, I suggest that you improve the figure to make it more explicit.
In addition, I noticed that a figure depicting the full pipeline is missing, as Figure 3 does not include the final classification using the MLP and the initial reconstruction of the cerebellar surface. Please provide a comprehensive figure to help readers understand the full pipeline.
It seems the code for this work is not provided and many details and hyperparameter to train the pipeline are missing. This could affect reproducibility of this work.
Minor: I noticed that there are several cases where you have started a sentence with a capital letter after a semicolon. Please ensure that you follow proper grammar format.
- 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?
Overall, I found the approach interesting, although it is a simple combination of pre-existing approaches (Graph CNN, existing contrastive loss, etc.). Validation and experimental setting seem limited but probably could be enaugh for this application.
Many details for the actual implementation are also missing and this could affect reproducibility.
- 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
In this paper, the authors proposed an automated method for anatomically meaningful cerebellar cortical surface parcellation.
- 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 paper propsoed a novel cortical surface-based analysis method of the cerebellar cortex.
- The experiment results aer promising
- The paper have several contributions and is generally clearly presented.
- The proposed methods look technically 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.
- Improve the organization of the papers.
- Improve explanation of the algorithm.
- Adding some new experiments.
- 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
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/2023/en/REVIEWER-GUIDELINES.html
See Q6
- 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 paper is well-written and the propsoed method is interseting
- 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
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.
- reviewer 1 mentions that 10 subjects is on a smaller side in terms of validation sample size. A larger sample would be more appropriate.
- Good idea for cortical parcellation using MLP
- Some methodological details pointed out by reviewer 2 are missingand should be added
Author Feedback
N/A