List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Claire Cury, Jean-Marie Batail, Julie Coloigner
Abstract
Major depression is a leading cause of disability due to its trend to recurrence and treatment resistance. Currently, there are no biomarkers which could potentially identify patients with risk of treatment resistance.
In this original paper, we propose a two-level shape analysis of the white matter bundles based on the Large Diffeomorphic Deformation Metric Mapping framework, to study treatment resistant depression. Fiber bundles are characterised via the deformation of their center line from a centroid shape. We developed two statistical analyses at a global and a local level to identify the most relevant bundles related to treatment resistant depression.
Using a prospective longitudinal cohort including 63 patients. We applied this approach at baseline on 50 white matter fiber-tracts, to predict the clinical improvement at 6 months. Our results show a strong association between three bundles and the clinical improvement 6 months after. More precisely, the right-sided thalamo-occipital fascicle and optic radiations are the most robust followed by the splenium. The present study shows the interest in considering white matter shape in the context of depression, contributing to improve our understanding of neurobiological process of treatment resistance depression.
Link to paper
DOI: https://link.springer.com/chapter/10.1007/978-3-031-16431-6_16
SharedIt: https://rdcu.be/cVD4Y
Link to the code repository
N/A
Link to the dataset(s)
N/A
Reviews
Review #2
- Please describe the contribution of the paper
The authors present an association of biomarkers created from shape of white matter fiber tracts with outcome in major depression in a longitudinal study of 63 individuals. The authors compute the shape metrics of the fibers from an MRI taken at start of the study and try to predict the outcome after 6 months, using their biomarkers and other relevant covariates.
- 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 describes a study, where the authors have defined a proper experiment. The description of methodology is mostly clear and the paper is generally well written. Technical aspects are described, all the way from getting ethical permissions for the study, image acquisition, preprocessing of data, major extraction of relevant features and statistical analysis.
The authors furthermore describe their findings.
- 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 has some weaknesses that could be addressed.
1) The paper assumes a lot of technical knowledge from the reader, and could sometimes benefit from being a bit more clear in explaining what exactly is meant by specific words. E.g. in the abstract the authors state “Shapes are characterised via the deformation of their center line from a centroid shape.” It is not clear at this point what is meant by “a shape”. A shape could be a path in 3d space, a surface, a volume, this needs to be clarified. If it is a path, then it can be a closed loop, non-closed, or something that branches, this needs to be a bit more clear, since the word shape is very generic.
2) The novelty seems to be in using these shape features for the following analysis. It is not exactly clear what has been done prior. I would also have wanted to see a comparison to some baseline, where other kind of information is extracted from the bundles.
3) The paper could explain the dataset better. E.g. depression is more common in women, is this reflected in the dataset? What was the depression severity of the patients at the start of the experiment? What is the age distribution? This should all be in a supplementary table, along with all the other covariates used for the linear models. Each covariate should also be described.
4) A figure could aid sections 3.1 and 3.2. This could also be used to clearly define what is meant by a shape. It seems to be a path defined by a smooth mapping from the unit interval [0,1] to R^3. This figure should ideally clearly explain the local and global features.
5) The analysis could be improved (see detailed comments).
6) I think it is very strong wording to call the findings a biomarker.
7) The size of the dataset (number of individuals) should be mentioned in the abstract and the introduction.
- 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
This report is consistent with the paper.
- 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
Specifics about analysis:
There are two major issues with the analysis. These both stem from the fact that this is a relatively small sample with many covariates. You do not have much statistical power to detect any associations, except for very strong effects. There are two things that you do that I believe are inflating the association results, and this is not a standard way to do association analysis.
1) I do not understand the point of adding noise to the bundles in section 3.3, it sounds like an attempt to do some regularization because of the excessive number of covariates compared to the number of samples. This is fine in machine learning and statistics, where you are solving a prediction problem, but this does not seem correct in the setting of doing association, it actually might result in double dipping and inflated association statistics.
2) I understand the point of doing leave one out cross-validation, but it makes more sense for prediction. In this setting, for the each of the 1000 models run, I would have wanted to see two models. One with only standard covariates and the other with the standard covariates and then also the ones specific to the particular bundle (local or global). Then the authors should perform a likelihood ratio test whether the model with the shape data is significantly better than the model that has only the standard covariates. This is a standard approach for this kind of analysis. The results seem to be the adjusted R squared on the case that is left out in cross-validation. This R squared is using the standard covariates with the shape data, so it doesn’t reflect only the contribution of the shape data, but also the standard covariates, and is thus very likely inflated and not truly representative of the association the researchers seem to be after.
Comments on English language:
1) This sentence in section 3.3 is wrong w.r.t. English language: “We did not added more landmarks…”.
2) It sounds weird at the start of section 3.3 to say that you “performed” two linear models. I would say that you either fit two models, or that you defined two models. Defined sounds more correct here, since you furthermore fit a bunch of times for cross-validation.
- 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?
This paper is generally very good, but I am not convinced by the approach used for doing the association analysis. If this is fixed, then I think this should be accepted, but I am concerned that this might be too much to ask for at this stage.
- Number of papers in your stack
5
- What is the ranking of this paper in your review stack?
3
- 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
6
- [Post rebuttal] Please justify your decision
The authors acknowledge my concerns and say that they will mention the limitations I point out in the manuscript. The authors also state that they will tone down the wording of the results, and not call it a biomarker.
Review #3
- Please describe the contribution of the paper
The paper is about major depressive disorder and white matter imaging. The methods describe a technique to predict clinical improvement at six months.
- 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 good, according to accepted standards
- 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 voxelsize is anisotropic; 2 x 1 x 1, which causes issues on the tensor model fitting.
- A specific result was found on the right hemisphere. There was no correction for handedness in the study. In order to be correct, this needs to be added to the model.
- 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
Should be reproducible based on what has been reported. Though, code is missing on how the methods were executed.
- 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
Interesting findings in the splenium of the corpus callosum, the right optic radiation and the right thalamo-occipital fascicule. How would this fit with major depressive disorder?
- 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
7
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The clinical application is interesting. Not only a novel technical application, but also working towards a biomarker.
- Number of papers in your stack
4
- What is the ranking of this paper in your review stack?
2
- Reviewer confidence
Somewhat Confident
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
Not Answered
- [Post rebuttal] Please justify your decision
Not Answered
Review #4
- Please describe the contribution of the paper
The article describes a two levels shape analysis (global and local) of fiber bundles based on the large diffeomorphic deformation metric to discriminate treatment resistant depression.
- 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.
Very few studies are focused on tractography as a biomarker for treatment resistant depression. As we still don’t have widely accepted imaging biomarkers for depression, this is study is a relevant contribution to the field. It is appreciable the cure into the neuroanatomy details of the results.
- 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 approach seems novel, to the knowledge of the review not many serious concerns. Mostly text improvements and capitalization in the bibliography.
- 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
They used publicly available tools. However, the dataset is not public, hopefully it will be. Used parameters are reported but they specific code is not available or not clear whether will be. I was not able to find the the reference foot note 1 to what is referred (the github link).
- 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 first sentence is a bit misleading. Depression is a wide spectrum of mental illness, some minor, some mild and some dangerous. Please specify that you are referring to major depression, or treatment resistance earlier. Riemannian is upper case Riemannian Diffusion image processing: please report which tools you used. Capitalization in bibliography: 7-t in bibliography should be 7-T or 7T, mri should be MRI, etc
- 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?
I have not been able to find critical issues.
- Number of papers in your stack
3
- What is the ranking of this paper in your review stack?
1
- Reviewer confidence
Somewhat Confident
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
Not Answered
- [Post rebuttal] Please justify your decision
Not Answered
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 work applies LDDMM framework to study connectivity changes in major depression. This is largely an application of existing methods, so the soundness of statistical analysis and biological relevance of the results are critical. While promising results were reported, there are concerns regarding the statistical analyses from reviewer 1. The biological relevance of the finding in the fiber bundles from corpus callosum and optic radiation should be justified and discussed.
- 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).
5
Author Feedback
We would like to thank reviewers for their insightful comments. We were delighted to read that all reviewers judged our submission worth being published. Reviewer 2 rightfully pointed out limitations in our analysis: 1) “I do not understand the point of adding noise to the bundles in section 3.3, it sounds like an attempt to do some regularization because of the excessive number of covariates compared to the number of samples. This is fine in machine learning and statistics, where you are solving a prediction problem, but this does not seem correct in the setting of doing association, it actually might result in double dipping and inflated association statistics” We acknowledge that adding noise to the bundles artificially reduces the p-values, and added this as a limitation in the new version of the paper. Furthermore, this relates to another comment (“I think it is very strong wording to call the findings a biomarker”) and we agree to use a lighter word and change biomarker with feature. We do not mention p-values in the paper, however for the bundles of interest we have: (CC7/OR/T-OCC) has for the global model a median p-value of (10e-11/10e-14/10e-18) and (10e-18/10e-16/10e-13) for the local model. 2) “I understand the point of doing leave one out cross-validation, but it makes more sense for prediction. In this setting, for the each of the 1000 models run, I would have wanted to see two models. One with only standard covariates and the other with the standard covariates and then also the ones specific to the particular bundle” This is a good point, we did not add a model with only standard covariates, as in all significant models, shape features were always significantly non null, and very often the strongest covariables. This means that shape features (globals or locals) do contribute to the model, otherwise null weight would be estimated. We now make that clearer on the paper. However, the proposed analysis will definitely be added in an extended version of this study. We will also upload the results excel file in supplementary material. 3) “The results seem to be the adjusted R squared on the case that is left out in cross-validation. This R squared is using the standard covariates with the shape data, so it doesn’t reflect only the contribution of the shape data, but also the standard covariates, and is thus very likely inflated and not truly representative of the association the researchers seem to be after.” The adjusted R square has lower values than the R square. Not sure to understand this point, as it does not “inflate”. A clue would be to contrast the adjusted R square with the model fitted with only standard covariates. However, here we chose to not consider those covariates, as they are just here to cancel effects of potentially linked variables and make sure not to reflect in the shape features, effects coming from standard covariates. Regarding covariates, a comment from reviewer 3 was “There was no correction for handedness in the study”. We eventually removed this covariate as there are only 2 left-handed patients out of 66 in the dataset. The asymmetry found in our analysis is not due to this lack. As reported in the paper, this asymmetry is also reported in other studies [3]. This information is provided as supplementary material, along with demographic info and other covariates as suggested by reviewer 2. Finally, reviewer 3 commented : “Interesting findings in the splenium of the corpus callosum, the right OR and the right T-OCC. How would this fit with major depressive disorder?” Widespread abnormalities in MDD have been reported [a] specifically in CC7 which has been linked with anxiety [5] and abnormalities of cortico-subcortical projections (such as in OR and T-OCC) are involved in cognitive and emotional regulation in depression [b,c]. New references: [a] Van Velzen, L. et al. Mol Psychiatry (2020). [b] Yan, B. et al. Front. Neurosci. (2020). [c] Long, Y. et al. NeuroImage Clin. (2020).
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 has sufficiently addressed reviewer questions, and there is a consensus on the technical merit of the proposed work.
- 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/n (best paper in your stack) and n/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).
8
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.
While this work is largely an application of the LDDMM framework for studying tract shapes, this is still a good contribution in terms of the clinical application. This was also supported by reviewers comments which unanimously accepted the paper.
- 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/n (best paper in your stack) and n/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).
8
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.
This submission largely relies on a combination of established methods, and a focus is on the analysis of a novel clinical dataset. I am a bit surprised to see such a work at MICCAI instead of a more clinically oriented journal. However, all reviewers see sufficient novelty to support acceptance to MICCAI. The rebuttal constructively addresses the reviewer concerns, openly discusses limitations, and I expect that authors will improve their final version accordingly.
- 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/n (best paper in your stack) and n/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).
3