List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Defu Yang, Hui Shen, Minghan Chen, Yitian Xue, Shuai Wang, Guorong Wu, Wentao Zhu
Abstract
Advancements in neuroimaging technology have made it possible to measure the connectivity evolution between different brain regions over time. Emerging evi-dence shows that some critical brain regions, known as hub nodes, play a signifi-cant role in updating brain network connectivity over time. However, current spa-tiotemporal hub identification is built on static network-based approaches, where hub regions are identified independently for each temporal brain network without considering their temporal consistency, and fails to align the evolution of hubs with changes in connectivity dynamics. To address this problem, we propose a novel spatiotemporal hub identification method that utilizes dynamic graph embedding to distinguish temporal hubs from peripheral nodes. Specially, to preserve the time consistency information, we put the dynamic graph embedding learning upon a smooth physics model of network-to-network evolution, which mathematically expresses as a total variation of dynamic graph embedding with respect to time. A novel Grassmannian manifold optimization scheme is further introduced to learn the embeddings accurately and capture the time-varying topology of brain network. Experimental results on real data demonstrate the highest temporal consistency in hub identification, surpassing conventional approaches.
Link to paper
DOI: https://doi.org/10.1007/978-3-031-43895-0_37
SharedIt: https://rdcu.be/dnwyQ
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 proposed a new method to identify the spatiotemporal hubs by considering temporal consistency. The new methods led to improved consistency in hub identification.
- 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 provides better temporal consistency during the hub identififcation.
- 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.
- Regarding the brain, there is no underground truth for hubs. Thus, it is more important to discuss what the hubs represent in neuroscience.
- It is important to clarify in what scenarios the consistency of hubs make a different. For example, the authors may discuss the roles of hubs in the dynamic functional connectivity reconfiguration and further examine their roles in support the execution of tasks or state transition.
- The comparison in Fig.4 is not significant and does not support the importance of the proposed method.
- Please rate the clarity and organization of this paper
Satisfactory
- 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 authors provided necessary information for the reproducibility.
- 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 check the weakness.
- 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
3
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
I scored the paper based on the insufficient validation of the proposed framework.
- 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
The contribution of the paper is the proposal of a novel spatiotemporal hub identification method that addresses the limitations of existing approaches for identifying hub nodes in dynamic scenarios. The proposed method utilizes dynamic graph embedding and a Grassmannian manifold optimization scheme to jointly identify a set of temporal hub nodes while ensuring their temporal consistency and alignment with the dynamic connectivity evolution in real time. The results on both simulated and real data demonstrate the effectiveness of the proposed method and its potential for investigating the role of hubs in the evolution of both task-based and resting-state-based networks.
- 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 main strength of the paper is the development of a novel spatiotemporal hub identification method for dynamic brain network analysis. The proposed method addresses the limitations of existing approaches by jointly identifying temporal hub nodes using dynamic graph embedding and leveraging a smooth physics model of network-to-network evolution to preserve time consistency information. The method is evaluated on both synthetic and real brain network data and outperforms conventional approaches in terms of temporal consistency of hub identification. The study shows technological and medical imaging novelty. It is well written and organised
- 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.
Limited evaluation: Although the proposed method is evaluated on both synthetic and real brain network data, the evaluation is limited to only a few metrics, and a more comprehensive evaluation that includes a comparison with other state-of-the-art approaches is lacking. Lack of real-world application: While the proposed method shows promising results, the paper does not discuss any potential real-world applications of the method in clinical settings or other domains. Lack of user-friendliness: The proposed method is highly technical and may be challenging for researchers without a strong background in mathematical optimization and dynamic graph embeddings to understand and implement.
- 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 paper is reproducable.
- 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
Thank you for submitting your paper on spatiotemporal hub identification in brain networks. Your proposed method for identifying temporal hub nodes using dynamic graph embedding and a Grassmannian manifold optimization scheme is an interesting approach to addressing the limitations of existing methods for dynamic scenarios. Overall, your study makes a valuable contribution to the field of neuroimaging.
However, there are several areas where the paper can be improved. Firstly, please consider to write more clear the pages 3-5 the mathematics as it is a bit difficult to read. This will make it easier for readers to understand and replicate your method.
Secondly, the evaluation of your proposed method could be further improved. While the results on both synthetic and real data are promising, there is a need for more detailed analysis and comparison with other state-of-the-art techniques. Specifically, it would be helpful to provide a more comprehensive comparison with existing approaches for dynamic hub identification, including a detailed analysis of the advantages and limitations of your method compared to others. This would provide readers with a better understanding of the strengths and weaknesses of your proposed approach.
Finally, the figures presented in the paper are difficult to interpret and do not provide a clear representation of the comparison between your proposed method and other approaches. The authors are encouraged to improve the clarity and quality of the figures. Please consider to reorganise Fig 1.
In summary, the proposed method for spatiotemporal hub identification in brain networks is an interesting contribution to the field of neuroimaging. However, the paper can be further improved by providing a more clear explanations of the methodological approach, a more thorough evaluation of the proposed method (or mention as future work please), and better quality figures.
- 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?
In summary, the proposed method for spatiotemporal hub identification in brain networks is an interesting contribution to the field of neuroimaging. However, the paper can be further improved by providing a more clear explanations of the methodological approach, a more thorough evaluation of the proposed method (or mention as future work please), and better quality in some figures (like Fig. 1 etc) .
- 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 #4
- Please describe the contribution of the paper
The paper extends a graph-embedding approach to detect regional hubs in static functional connectivity to detect the hubs by considering the dynamic functional connectivity patterns. The hub detection is posed as one-shot constrained optimization problem. The potential of the method is demonstrated on a synthetic dataset and one resting-state fMRI 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.
The paper extends a graph-embedding approach to detect regional hubs in static functional connectivity to detect the hubs by considering the dynamic functional connectivity patterns. The hub detection is posed as one-shot constrained optimization problem in a spatio-temporal framework rather that making inferences on a temporal sequence of hubs.
- 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 paper is difficult to understand without reading through the references - for example, how the graph-embedding is achieved and the equation (1).
- The optimization scheme is not novel as authors claim.
- Since the ground truth hubs are unknown, it is difficult determine whether getting more hub nodes in real data shows the superiority of the method.
- Details on how synthetic data was created is scanty.
- 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 implementation and reproducibility of the methods would be difficult without the access to the codes.
- 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
The paper extends a graph-embedding approach to detect regional hubs in static functional connectivity to detect the hubs by considering the dynamic functional connectivity patterns. The hub detection is posed as one-shot constrained optimization problem in a spatio-temporal framework rather that making inferences on a temporal sequence of hubs.
- The paper is difficult to understand without reading through the references - for example, how the graph-embedding is achieved and the equation (1).
- The optimization scheme is not novel as authors claim.
- Since the ground truth hubs are unknown, it is difficult determine whether getting more hub nodes in real data shows the superiority of the method.
- Details on how synthetic data was created is scanty.
- The implementation and reproducibility of the methods would be difficult without the access to the codes.
- 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?
Though the methodology looks promising, details of methodology, experiments, and results are scanty.
- 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.
This paper proposed a spatiotemporal hub identification method to improve temporal consistency. It was evaluated on both synthetic and real brain network data. There are also some concerns. As reviewers pointed out, the benefit on real-world application is not well described, evaluations could be improved.
Author Feedback
We appreciate the reviewers’ comments on the novelty of our work. Their major concerns are 1) Validation; 2) Real world application; 3) Explanation of the method. We first reply to these questions and then answer specific questions from the reviewers.
Validation
We agree with the reviewers’ comments that there is no ground truth in real brain networks. Thus, we validate our proposed approaches based on the synthesized and the real brain networks. Firstly, we use the synthesized data to validate the accuracy and the robustness. Then, we indirectly evaluate the performance according to the following two hypotheses in neuroscience: 1) hubs are stable regardless of brain states switching from task-to-task and play the role of network reconfiguration (verified the consistency of hub in Fig.4); 2) Selective network vulnerability-neuropathological burdens selectively affect hub regions (verified the quality of hubs associated with neuro-disease in Fig.5). These hypotheses are currently mutual consent in the neuroscience community. In terms of the accuracy, most of the hubs identified by our proposed method are frequently reported as the major hallmark in the progression of OCD. Moreover, due to the limitation of the paper length, we selected two of the most representative methods as references: 1) Sorting-based by using handcrafted graph centrality (Betweenness); 2) Learning-based (Graph-embedding). We report the results obtained by the above two representative methods since they performed the best among the various state-of-the-art approaches in our experiments. As the reviewers suggest, a more comprehensive evaluation will be conducted in our future work.
Real world application
One of the most crucial motivation to propose the spatiotemporal hub identification is to provide a potential tool in real world applications, such as the evolution of hubs in supporting the execution of task and state transition. By applying our proposed method on working memory task-based fMRI datasets, the results of the brain states switch are quite promising, showing that the average task-to-task (consist of eight task states) similarity of temporal hubs is larger than 0.997 (0.972 of Sorting-Betweenness-based, and 0.984 of Graph-embedding-based methods). In our future work, additional experiments will be conducted to investigate the role of hub on connectivity reconfiguration and examine their role in supporting the execution of tasks.
Explanation
The fundamental of Eq. (1) is to find a set of nodes whose removal would result in a maximum number of disconnected components in the network, where the number of disconnected components is equal to the number of zero eigenvalues of the Laplacian matrix. We solve the optimization problem “maximize the number of zero eigenvalues” with an approximation: “minimize the summation of eigenvalues”. Since the topology governed by Laplacian matrix is often represented as a graph embedding, spatiotemporal hub identification can be easily casted as a dynamic graph embedding learning problem. After publication of our work, our code and a demo will be released.
R1
Thanks. We will add some crucial discussions. One of the most representative promising results has been reported in the reply of “Real world application”. Q3 The comparison in Fig.4 is not significant and does not support the importance of the proposed method. -: The reason is that these three temporal networks have a relatively large window size of 50 seconds, which leads to the homogeneity of temporal networks. When reducing the window size, the performance of our proposed method will become more significant than the alternative methods. The smaller window size, the more dynamics. For example, by cutting the entire time series into 7 (20s) and 16 (10s) temporal networks, the average results are 0.90 and 0.93 by our methods, 0.50 and 0.22 by Sorting-Betweenness-based, 0.73 and 0.54 by Graph-embedding-based methods, respectively.
Post-rebuttal Meta-Reviews
Meta-review # 1 (Primary)
- Please provide your assessment of the paper taking all information into account, including rebuttal. Highlight the key strengths and weaknesses of the paper, clarify how you reconciled contrasting review comments and scores, indicate if concerns were successfully addressed in the rebuttal, and provide a clear justification of your decision. If you disagree with some of the (meta)reviewer statements, you can indicate so in your meta-review. Please make sure that the authors, program chairs, and the public can understand the reason for your decision.
The authors proposed a graph-embedding approach for dynamic functional connectivity patterns of hubs. Concerns of validation and real world appliication were addressed in the rebuttal.
Meta-review #2
- Please provide your assessment of the paper taking all information into account, including rebuttal. Highlight the key strengths and weaknesses of the paper, clarify how you reconciled contrasting review comments and scores, indicate if concerns were successfully addressed in the rebuttal, and provide a clear justification of your decision. If you disagree with some of the (meta)reviewer statements, you can indicate so in your meta-review. Please make sure that the authors, program chairs, and the public can understand the reason for your decision.
This paper introduces a new method to identify the spatiotemporal hubs by considering temporal consistency. Reviewers have concerns with the unclear writing of the current version of paper. It might need major revision of the draft. Also, evaluations are not strong to support the main claim of the paper. It would be great to improve the paper.
Meta-review #3
- Please provide your assessment of the paper taking all information into account, including rebuttal. Highlight the key strengths and weaknesses of the paper, clarify how you reconciled contrasting review comments and scores, indicate if concerns were successfully addressed in the rebuttal, and provide a clear justification of your decision. If you disagree with some of the (meta)reviewer statements, you can indicate so in your meta-review. Please make sure that the authors, program chairs, and the public can understand the reason for your decision.
The paper proposes a graph embedding to model brain hub. Due to lack of ground truth they validated their method on synthetic data. For the real data, they relied on hub stability hypothesis in neuroscience. The paper is acceptable although falsiablity is a major issue in such studies.