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

Authors

Jiong Zhang, Yonggang Shi

Abstract

The temporal cortex is one of the earliest regions with tau pathology and associated gray matter atrophy in Alzheimer’s disease (AD). Surface mapping has conventionally been widely used to provide one-to-one correspondences and hence the detection of thickness changes in the cortical ribbon. The presence of very different topography of the sulcal and gyral folds across subjects, however, makes it challenging to have meaningful sulcus-to-sulcus and gyrus-to-gyrus matching. This is critical for the quantification of thickness changes because sulcal and gyral areas have different thickness profiles. In this paper, we propose a novel framework for personalized and localized cortical folding pattern analysis to address this challenge. Given a pair of source and target patches, intrinsic surface mapping based on Riemannian metric optimization on surfaces (RMOS) is first employed to compute the fine-grained maps. Afterwards, we design an edge-distortion based pattern matching method to detect locally well-matched folding patterns between temporal cortical patches. A patch-based similarity measure is then defined to establish a personalized atlas set for each individual source patch. Finally, a personalized z-score map is computed for normality assessment in disease groups and the detection of atrophy with respect to the normal controls. The proposed framework is validated on a large-scale dataset from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to demonstrate the effectiveness of the proposed framework for personalized analysis and increased power in the detection of atrophy in AD and mild cognitive impairment (MCI).

Link to paper

DOI: https://doi.org/10.1007/978-3-030-87234-2_67

SharedIt: https://rdcu.be/cyl9d

Link to the code repository

N/A

Link to the dataset(s)

N/A

Reviews

Review #1

Please describe the contribution of the paper

This paper proposed a method to match cortical patches to one of least distorted atlases, and only use regions with strong similarity to atlas for statistical comparison.
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.

Multi-atlas and local similarity level is included during brain mapping, which might help to increase statistical significance.
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. using thickness instead of curvature for mapping feature might effect further analysis of thickness.
2. Though pattern matching seems increased Z-score on AD and MCI individual, it might also increased Z-score of NC.
3. This method may be hard to use, since the similar and dissimilar region may change subject by subject.
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 data is public available.
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 Add NC (e.g. 150 not used in atlases) to compare in the results (Fig 5).
1. Validate if use curvature instead of thickness during mapping.
Please state your overall opinion of the paper

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

The NC subjects should validated in the results, otherwise we don’t know if the Z-score is systematically increased.
What is the ranking of this paper in your review stack?

3
Number of papers in your stack

4
Reviewer confidence

Confident but not absolutely certain

Review #2

Please describe the contribution of the paper

Alzheimer’s disease begin with affecting the hippocampal complex by building intracellular plaques that lead to cell death, which can be measured as a cortical thinning in MRI. This submission describes a method for matching MRI-derived surface patches derived within a cohort. Because it is well-known that cortical thickness differs between gyri and sulci, it is important to maintain their relationship in the matching process. A curvature-based method (Riemannian metric optimization on surfaces, RMOS) is employed here. Results obtained in a large, publicly available data base (ADNI) demonstrate an advance in precision when maintaining the sulcal/gyral reference frame.
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 topic of this manuscript is within the scope of this conference, and of potential interest to its audience. The text is well-written, without major errors (except where noted below), and readily understandable for a reader with a moderate background in medical image analysis.
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 RMOS approach was published before, and is applied here slightly modified to a different problem, with an added standard statistical analysis. While there is “not much news” here, the work reported here is sound and advances the analytic technology in this field.
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

“An analysis of situations in which the method failed.” - “Not applicable”: true?
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
Two minor issues should be taken care of:
1. p.3: “…we have shown that large … variations … are existed in the temporal region.” Please, rephrase.
2. p.5: “…we can observe the large folding pattern variations” -> “we can observe large folding pattern variations”
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?

Little news, but solid work.
What is the ranking of this paper in your review stack?

3
Number of papers in your stack

5
Reviewer confidence

Very confident

Review #3

Please describe the contribution of the paper

This paper proposed a novel idea of quantifying the abnormality of cortical thickness in MCI and AD patients by looking at the deviation from a set of control individuals with similar folding patterns based on distortion ratio derived from the Riemannian metric optimization on surfaces (RMOS) method. The proposed approach points out a novel direction of quantifying thickness differences, which will promote interesting discussions.
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.

• This is a technical sound paper. Although the idea of accounting for the confound of anatomical variability of the cortical folding pattern is not new but building a personalized atlas set with subjects having a similar folding pattern and quantify the thickness abnormality is a novel direction to investigate. • The evaluation is comprehensive and supports the conclusions of this paper.
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.

• Overall, this paper is well-written and well-organized. One weakness is that section 2.2 is hard to follow that might be due to (1) too many math symbols together with subscripts and superscripts were introduced at the same time, (2) lack of introduction of some concepts, e.g. LB embedding, (3) the formulation of E_R was missing. It would be helpful if the authors can provide some more detailed introduction in the supplementary material. • The clinical feasibility of the proposed approach was not evaluated, i.e. the statistical comparison between controls and disease groups (MCI/AD) using the proposed approach was missing. This would be a great addition to this paper
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

Among the applicable items the authors selected in the reproducibility checklist, I have the following comments. (1) Item 1.2: The study cohort was not discussed in detail as no characteristics of the cohort were provided (age, gender, education, cognitive test scores, how they were selected and so on). (2) Item 3.1 and 3.2: No information was provided about the values of hyper-parameter and how they were tuned as well as how sensitive were the results to the hyper-parameters.
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

In addition to the above comments, please find below some minor points: • Some abbreviations were not defined: NC, LB, ADNI. • In the third paragraph of the introduction, it would be nice to provide some citations to the prior work on sub-typing the cortical folding patterns. • The results in Figure 4 from 3 randomly selected cases are nice but may not be convincing as they may vary from subject to subject. It would be nice if the authors can also include box-plots for the whole NC group (each subject provide two mean thickness differences in high similarly and low similarly groups).
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?

• As indicated above, the proposed paper started a new direction of quantifying thickness differences, which will promote interesting discussions.
What is the ranking of this paper in your review stack?

1
Number of papers in your stack

3
Reviewer confidence

Confident but not absolutely certain

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 proposes personalized matching and analysis cortical folding patterns for addressing large inter-subject variability of folding patterns for better studying the cortical thickness differences betwwen normal and AD subjects. Major strengths include important and promising reseach direction, plausibile method, clear presentation, and solid experiments. Major weaknesses: 1) Using cortical thickness, which is usually noisy and less reliable compared to folding features, for mapping cortical surfaces might adversely affect the mapping/registration accuracy, thus further affecting analysis of thickness. 2) Although pattern matching seems increasing Z-score on AD and MCI individuals, it might also increase Z-score of NC. 3) This method seems hard to use in practice, since the similar/dissimilar regions may change subject by subject and region by region. 4) The RMOS approach was published before, and is applied here slightly modified to a different problem.
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).

2

Author Feedback

We thank the reviewers and area chair for their positive comments about the novel contribution of opening important and promising research direction by our personalized cortical folding pattern matching and analysis framework. Here we clarify the main questions raised by the reviewers. For questions, we indicate the reviewers as R1, R2, R3, and RM (meta-review), respectively.

Selection of mapping feature. (RM&R1) We appreciate the reviewers’ suggestions for choosing stable features for mapping. In the current feature selection, we use cortical thickness as an initial attempt to guide the intrinsic surface mapping process to get anatomically meaningful sulcus-to-sulcus and gyrus-to-gyrus local correspondences. By giving a few iterations, our analysis shows that reasonable temporal pattern matchings can be achieved. Our main point here is to show the potential of further exploring personalized analysis by taking surface mapping as an initial step. The experimental validations show the benefits of the proposed similarity matching concept for obtaining a personalized atlas set for individual. We also tested other features such as local curvature in our previous experiments and they all achieve similar performances. Nevertheless, we agree with the reviewers that more prominent features can lead to better mapping accuracy. In our future investigations, we will employ better feature functions to guide more precise local pattern matchings for our personalized analysis.

Pattern matching also increases Z-score of NC. (RM&R1) The goal of performing personalized pattern matching and analysis is to detect localized and subtle pathological changes that has been obscured in standard mappings. Thus, the personalized normality will become more significant for all groups and it is not surprising that the Z-score will also increase for some of the NC subjects with respect to their NC atlas distributions. However, since the pathological brain atrophy changes are more significant,the proposed framework is more powerful at detecting such changes due to disease progression and can obtain higher deviation from the NC atlas. This will give better discrimination ability for the early detection and classification of AD.

Hard to use in practice, since the similar/dissimilar regions may change subject by subject and region by region. (RM&R1) The proposed concept is not hard to implement in practice, and the main purpose of the proposed framework is exactly to detect personalized atrophy patterns that are different from subjects and hence reduce errors caused by the local pattern variations across subjects. Our method mainly includes three steps. (1) A surface mapping module is employed as an initialization step to achieve good point-wise correspondence between patches. (2) A predefined vertex-wise similarity measure is used to obtain the vertex-wise atlas on the surface. (3) A vertex-wise normality score is computed with respect to its own atlas and thus a personalized normality map can be obtained for each subject.

Novelty/contribution of this work. (RM&R2) The main novelty/contribution of this work is that a personalized cortical folding pattern analysis framework is proposed to solve the practical challenges of large cortical folding pattern variations in atrophy detection. Standard mapping techniques can hardly solve such issues since perfect sulcus-to-sulcus and gyrus-to-gyrus matchings do not exist in practice due to the large folding pattern variations. In this work, RMOS is only employed as a mapping tool for generating the initial point-wise correspondences. A more precise vertex-wise similarity measure is designed to detect similar local patterns around each vertex. Then a vertex-wise atlas set is constructed for localized analysis rather than the inaccurate analysis based on standard mapping. The proposed concept opens an important and promising research direction for analyzing personalized cortical surface changes.

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 rebuttal justified the contribution, slection of features and some results.
After you have reviewed the rebuttal, please provide your final rating based on all reviews and the authors’ rebuttal.

Accept
What is the rank of this paper among all your rebuttal papers? Use a number between 1 (best paper in your stack) and n (worst paper in your stack of n papers).

4

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.

The four major weaknesses 1), 2) and 4) are well addressed. 1) I agree with the authors’ response that cortical thickness could be one of the feature choices; 2)more pronounced increase of z-score on patients give better discrimination ability; 4) the work does not stop at applying RMOS but designs a vertex-wise similarity measure. While the authors did not provide a satisfied reply to the use in practice and the validity of the measures and methods were not fully presented, this work demonstrates a promising approach of vertex-wise folding comparative analysis.
After you have reviewed the rebuttal, please provide your final rating based on all reviews and the authors’ rebuttal.

Accept
What is the rank of this paper among all your rebuttal papers? Use a number between 1 (best paper in your stack) and n (worst paper in your stack of n papers).

1

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 simple cortical folding matching pipeline, that is quite intuitive, where similarly matched areas can be used for abnormality detection. Two out of 3 of the reviewers were positive about the paper. The results of the experimental validation are encouraging, and the application is valuable and of wide interest to the neuroscience imaging community. However, I am concerned about the lack of benchmarks. There is a large pool of cortical matching methods including spectral surface matching that should be stated and benchmarked against. A section on related works on cortical surface matching can be included in the final version and how RMOS stands out with respect to those. Given the clarifications that the rebuttal provided and the ranking of this paper in my pool, I recommend an accept.
After you have reviewed the rebuttal, please provide your final rating based on all reviews and the authors’ rebuttal.

Accept
What is the rank of this paper among all your rebuttal papers? Use a number between 1 (best paper in your stack) and n (worst paper in your stack of n papers).

1

back to top

Personalized Matching and Analysis of Cortical Folding Patterns via Patch-Based Intrinsic Brain Mapping