Paper Info Reviews Meta-Review Author Feedback Post-rebuttal Meta-Reviews

Authors

Yanfu Zhang, Liang Zhan, Shandong Wu, Paul Thompson, Heng Huang

Abstract

Diffusion MRI-derived brain structural connectomes or brain networks are widely used in brain research. However, constructing brain networks is highly dependent on various tractography algorithms, which leads to difficulties in deciding the optimal view concerning downstream analysis. In this paper, we propose to learn a unified representation for multi-view brain networks. Particularly, we expect the learned representations to convey the information from different views fairly and in a disentangled sense. We achieve the disentanglement via an approach using unsupervised variational graph auto-encoders. We achieve the view-wise fairness, i.e. proportionality, via an alternative training routine. More specifically, we construct an analogy between training the deep network and the network flow problem. Based on the analogy, the fair representations learning is attained via a network scheduling algorithm aware of proportionality. The experimental results demonstrate that the learned representations fit various downstream tasks well. They also show that the proposed approach effectively preserves the proportionality.

Link to paper

DOI: https://doi.org/10.1007/978-3-030-87234-2_48

SharedIt: https://rdcu.be/cyl8K

Link to the code repository

N/A

Link to the dataset(s)

N/A

Reviews

Review #1

Please describe the contribution of the paper

In this paper, the authors present an unsupervised approach to learn unified graph embeddings for multi-view brain networks. As such, Graph Variational Auto-encoder has been utilized to learn the representations with disentanglement and proportionality. Further, they have performed classification for Parkinson’s and Alzheimer’s datasets.
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.

Overall the paper is written and presented properly.

The methodology is well explained and understandable.

The results are reproducible.

The author has also performed the ablation study.
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.

There is a limited amount of novelty in the work as they have used the existing deep learning technique, i.e., variational graph Auto-encoder.

Features visualization of the network is missing, which is essential for showing the learned distinguishable features of normal and patient.

Why have authors chosen three graph convolutional layers for the encoder part? They need to justify with proper reasons.

The amount of data over which GVAE has been trained very less. How authors overruled overfitting?

The comparative results shown in Tables 1, 2 are very close. With less data it’s statistical relevance is questionable.
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

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://miccai2021.org/en/REVIEWER-GUIDELINES.html

Various graphs like area under curve could be shown in quantitative analysis.

Besides, several other classification parameters like precision, recall, etc., can be reported.

Data agumentation can also be done.
Please state your overall opinion of the paper

borderline accept (6)
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

Limited novelty.
What is the ranking of this paper in your review stack?

3
Number of papers in your stack

5
Reviewer confidence

Confident but not absolutely certain

Review #2

Please describe the contribution of the paper

Based on the multi-view graph convolutional network, the authors attempted to extend the generalization ability of the newly proposed unified representation learning method into classifying both Alzheimer’s Disease (AD) and Parkinson’s Disease (PD). The accuracy of the proposed method was elevated when adding the representations of disentanglement and proportionality.
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.

Based on the multi-view graph convolutional network, the authors attempted to extend the generalization ability of the newly proposed unified representation learning method into classifying both Alzheimer’s Disease (AD) and Parkinson’s Disease (PD). The accuracy of the proposed method was elevated when adding the representations of disentanglement and proportionality.
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 general prediction of neurodegenerative conditions is not a major concern in clinical considerations. Furthermore, Table 1 shows the classification accuracy of AD (ADNI and NACC) was much higher than that of PD (PPMI), which seemed contradictory to the original purpose of general prediction.
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

Based on the multi-view graph convolutional network, the authors attempted to extend the generalization ability of the newly proposed unified representation learning method into classifying both Alzheimer’s Disease (AD) and Parkinson’s Disease (PD). The accuracy of the proposed method was elevated when adding the representations of disentanglement and proportionality.
Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://miccai2021.org/en/REVIEWER-GUIDELINES.html

The authors stated that the prediction ability was complicated in respect to different tasks and heavily coupled with the algorithms. For example, the key representations of disentanglement and proportionality indeed enhanced accuracy and minimized the error (though its efficacy depends on the tasks); however, the corresponding neurophysiological evidence shall be stated.
Please state your overall opinion of the paper

strong accept (9)
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

the authors attempted to extend the generalization ability of the newly proposed unified representation learning method into classifying both Alzheimer’s Disease (AD) and Parkinson’s Disease (PD). In clinical practice, we could try to use the methods for predication of AD and PD in different datasets
What is the ranking of this paper in your review stack?

3
Number of papers in your stack

3
Reviewer confidence

Confident but not absolutely certain

Review #3

Please describe the contribution of the paper

This work present a method to merge different connectome matrices information which can be used in downstream analysis.
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 loss proposed in this work is novel. It balances the proportionality of each kind of connectome.
2. The evaluation part which is performed on different classification datasets is fairly strong.
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.
1. It would be better if the authors can show the dimension of variables (\sigma and \mu) to help my understanding. 2.To my understanding, the information extracted from multi-view connectome is just a feature map. Would it be possible to use this method to generate a refined connectome matrix which can be regarded as a weighted sum of multi-view connectomes.
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

The reproducibility is not very good since a lot of details are missing such as input data size and intermediate feature map size.
Please provide detailed and constructive comments for the authors. Please also refer to our Reviewer’s guide on what makes a good review: https://miccai2021.org/en/REVIEWER-GUIDELINES.html
1. The figure quality could be improved and more details can be added to improve understanding. The dimension of the matrix variables frequently used in the paper should be given.
2. A important experiment setting which compares disentanglement network with original network is not given. My concern is that if the performance improvement is really brought up by disentanglement rather than the use of multiple DNNs.
Please state your overall opinion of the paper

borderline accept (6)
Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

The score is given based on the experimental setting and methodology of this work.
What is the ranking of this paper in your review stack?

2
Number of papers in your stack

5
Reviewer confidence

Very confident

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.

Although the paper is not highly innovative on a methodological level, the reviewers came to consensus that the paper has undeniable merits. We refer the authors to the constructive feedback provided by the reviewers. Please address their comments when revising the paper.
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).

3

Author Feedback

N/A

back to top

Disentangled and Proportional Representation Learning for Multi-View Brain Connectomes