List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Lipeng Ning
Abstract
Combinedrelaxation-diffusionMRI(rdMRI)techniqueprobes tissue microstructure using imaging data with multiple b-values and echo-time. Joint analysis of rdMRI data and provide the relaxation diffusion distribution (RDD) function to characterize heterogeneous tissue microstructure without using multi-component models. This paper shows that the problem of estimating RDD functions is equivalent to the multivariate Hausdorff moment problem by changing variables. Then three maximum entropy (ME) estimation problems are proposed to estimate the ME-RDD functions in different parameter spaces. All three problems can be solved by using convex optimization algorithms. The performance of the proposed algorithms is compared with the standard methods using basis functions based on simulation and in vivo rdMRI data. The proposed ME-RDD functions can provide more accurate estimation results.
Link to paper
DOI: https://doi.org/10.1007/978-3-031-43993-3_43
SharedIt: https://rdcu.be/dnwNO
Link to the code repository
https://github.com/LipengNing/ME-RDD
Link to the dataset(s)
https://github.com/LipengNing/ME-RDD/tree/main/MICCAI_example
Reviews
Review #2
- Please describe the contribution of the paper
The paper describes an equivalence between estimating the RDD (relaxation diffusion distribution) and the multivariate Hausdorff moment problem. 3 Maximum entropy estimation problems are presented to estimate maximum entropy RDD. Results are shown in simulated functions and in-vivo rdMRI data.
- 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.
A strength is the thorough theoretical deviation presented, guiding through all steps and variables in a very clear building-up way.
- 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 visualizations and figures are not so useful to appreciate the differences, the results shown and the graphics could do with some additional thought! The paper could also have done with another thorough read - lots of little errors in many sentences.
- 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
no specific effort for reproducibility can be seen. Neither data or any tools will be made 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://conferences.miccai.org/2023/en/REVIEWER-GUIDELINES.html
Detailed comments: Introduction -thorough description of the state of the art and good motivation Methods -see above (main strengths), clear derivation
- 3.1 - why where theses specific mean’s chosen? Would be good to point out to the reader what these combinations are meant to represent. -Same for SNR - would be good if an intuition could be given what these roughly represent (eg in which resolution / field strength etc they would be realistic?)
-For Figure 1 - less abbreviations in the figure would make it a lot more readable! eg there is space to write center of mass of D etc. -Caption could also include “for different simulated SNR levels” or similar.
-Lots of little errors eg repeated words (see for example page 6 last sentence “The data ACQUIRE ACQUIRED from healthy..”) - worth giving it another thorough read!
-Figure 2 - the visualization is difficult to appreciate the little differences we could observe. wondering if possible to show the D-R plan in 2D with a different scale?
-“Then the data was further processed using unring and MPPCA denoising tools” -unring needs a reference?
4 Results: -“and similar performance.” - similar to what? To ech other or to the L2 approach?
-“Fig. 3 shows the comparison of the ME-RDD (using ME-1) and L2-RDD in three type of tissue of human brains.” - would read simpler if just said in white matter, cortical and ubcortical gray matter. -Figure 3 is also quite hard to see! Given that this is all about comparisons really between L2 and ME1 in this case - why not making a difference plot or similar? scale is suboptimal and color scale missing.
Discussion -“The computation time to estimate the ME-RDD for one voxel is about 100 seconds using an Intel Xeon E5 cpu (2.20GHz).” This time could be in the results section.
- 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
4
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
My main reason to select “weak reject” is the quality of the results presentation and that it is quite difficult for the reader to appreciate the comparisons made.
- Reviewer confidence
Somewhat 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 paper shows the equivalence between estimating the relaxation diffusion distribution (RDD) in diffusion MRI with multiple echo times and b-values and the Hausdorff moment problem. Based on this equivalence, the authors apply the maximum entropy solutions for the Hausdorff moment problem to the estimation of the RDD. Simulation results show empirically that the solution based on maximum entropy is closer to the ground truth compared with the conventional basis-function approach.
- 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.
Simplicity: the formulation is interesting and easy to follow. Possible extensions: I think the connection of the third ME problem with minimizing the KL divergence is interesting. It would be nice to see how this minimization problem relates to the basis function representation. Presentation: the paper is written in a way that makes it easy to follow.
- 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.
Motivation: while ME is a well know approach for estimating distribution with moment constraints, the motivations for this particular approach is not very clear. For instance there is no discussion about what properties of ME solutions are desirable for the RDD problem.
- 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 paper lacks details on the implementation of the optimization algorithms. There was no converge criterion discussed or number of iteration or any hyper parameters. This makes the results hard to reproduce.
- 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 think a more clear motivation on how the ME approach can be better than the conventional approach would be valuable. Also if the authors can point out at when and how the standard basis function approach fails (not just form the experiments) can strengthen the motivation of the proposed work. It would be also good to discuss how the 3 different ME problem reach their solution and why they should differ. I look forward to future work with validated results for real data experiments. I also think the connection with the minimization with KL divergence is interesting and would like to see if the authors can elaborate on this part and perhaps connect it with the basis function approach. Minor comments: there are a couple of typos and missing and extra words in a few places including the abstract.
- 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 formulation is clear and does show to work better (at least empirically). I think the technique does open another avenue for work on the RDD estimation and it is worth discussing
- 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 #4
- Please describe the contribution of the paper
The present study proposes a novel approach for estimating the Radon transform of a probability distribution function, known as the relaxation diffusion distribution (RDD) function. Specifically, they introduce the maximum entropy estimation method to achieve more accurate estimation of the RDD function.
- 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 article has a coherent language and logical flow, concise formula derivation, and reasonable experimental design
- 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 specific motivation for using the ME method for improvement is unclear. And the method lacks experiments to see if it can help improve clinical diagnosis
- 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 of this paper is not clear.
- 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
1) The “Amijo” should be “Armijo”. 2) How did you determine the signal-to-noise ratio and related parameters for the synthetic data? 3) Is it possible to quantify Figure 2? 4) This paper lacks clinical experiments before and after methodological improvements
- 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 a decent paper. The rating is based on the good writing and experiments design. However the clinical value of this work is unclear.
- Reviewer confidence
Somewhat 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
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.
Reviewers were quite positive about this work on maximum entropy estimation of the relaxation diffusion distribution (RDD) function. In particular, they appreciated the thorough theoretical derivation and commented that it opens up an interesting avenue for RDD estimation. Reservations primarily related to aspects of the presentation, to the fact that reproducibility is limited (details are missing in the description, and neither code nor data will be available), and an unclear motivation for using an ME approach in this particular context, as pointed out by R3 and R4. In the rebuttal, authors should describe how they plan to address those concerns in their final version.
Author Feedback
We appreciate that all the reviewers understood the contribution of this paper in developing a maximum-entropy framework for estimating the joint relaxation and diffusion distribution. We thank all the constructive comments from the reviewers that help improve the quality of the paper. Reviewer #2 provided detailed comments about the text and suggestions to improve the figures. Reviewer #3 was very positive about this work and requested more discussion about the motivations and insights about the maximum entropy method. Review #4 ranked our work the 1 st out of 5 papers and requested more motivation and reproducibility.
We have carefully revised the paper following the review comments. Below are major changes in the revised version.
Response to Review #2: 1) As suggested, figure 1 was revised with new label names and a new caption.
2) The scale in Figure 2 was revised to show the differences between the methods better.
3) Figure 3 was revised to 2D images with the scale adjusted to show the differences.
4) The rationale for the experimental parameters and the SNR are added in Section 3.1. The TE and b-values are parameters that can be achieved using an advanced MRI scanner, e.g. the connectom scanner, for in vivo human brains. The SNR is within typical ranges of direction-averaged dMRI signals.
5) The typos and incorrect sentences are revised.
6)The code and data used in this paper will be shared via Github, which is added in the paper.
Response to Review #3 1) Regarding the maximum-entropy method, our motivation was based on the maximum-entropy power spectral estimation method in time series analysis. Maximum-entropy spectral estimation methods provide better higher resolution and more robust performance compared to Fourier transform-based methods with fewer samples. In the problem of estimation, RDD functions are related to a similar inverse problem as the spectral estimation with basis function-based method analogous to the Fourier-transform-based method. Thus, the proposed maximum-entropy RDD functions may perform better than basis-function methods, which is supported by the experimental results. The Kullback-Leibler (KL) is also a common approach for power spectral estimation. Our work showed an interesting relationship between the KL divergence and entropy function using a change of variable. We added discussions in the Introduction and Summary about the motivations.
2) The typos are corrected.
Response to Review #4 1) See response 1) to Reviewer #3 for the motivations of the maximum-entropy method.
2) Regarding reproducibility, the code and data used in this paper will be shared via Github. 3) The SNR of the synthetic data is based on the range of direction-averaged dMRI signals. The SNR of direction-averaged dMRI scales according to the square root of the number of gradient directions. The error metric for Fig. 2 is shown by the “noise-free” results in Fig.1. The caption of Fig. 2 is revised by adding “noise-free.” 4) Clinical application will be explored in future work. One sentence is added in the Summary about future work on applying the proposed method in lesion or tumor detections.
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.
Reviewers were rather positive about this work to begin with. The rebuttal is constructive, motivating the maximum-entropy approach in more detail and indicating that many issues regarding presentation will be improved in the final version. With respect to reproducibility, authors clarify that they will make their code publicly available. After the revisions described in the rebuttal, this should be a strong MICCAI contribution.
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 authors have adequately addressed the comments regarding the presentation, the reproducibility, unclear motivation, etc. They have noted that the reviewers’ comments will be incorporated into the final version of the paper. Hence, I suggest accepting this paper for MICCAI.
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 authors addressed all concerns summarized by reviewers (especially the motivation part). Therefore, I would recommend accept.