List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Kelly Payette, Alena Uus, Jordina Aviles Verdera, Carla Avena Zampieri, Megan Hall, Lisa Story, Maria Deprez, Mary A. Rutherford, Joseph V. Hajnal, Sébastien Ourselin, Raphael Tomi-Tricot, Jana Hutter
Abstract
Fetal Magnetic Resonance Imaging at low field strengths is emerging as an exciting direction in perinatal health. Clinical low field (0.55T) scanners are beneficial for fetal imaging due to their reduced susceptibility-induced artefacts, increased T2* values, and wider bore (widening access for the increasingly obese pregnant population). However, the lack of standard automated image processing tools such as segmentation and reconstruction hampers wider clinical use. In this study, we introduce a semi-automatic pipeline using quantitative MRI for the fetal body at low field strength resulting in fast and detailed quantitative T2* relaxometry analysis of all major fetal body organs. Multi-echo dynamic sequences of the fetal body were acquired and reconstructed into a single high-resolution volume using deformable slice-to-volume reconstruction, generating both structural and quantitative T2* 3D volumes. A neural network trained using a semi-supervised approach was created to automatically segment these fetal body 3D volumes into ten different organs (resulting in dice values > 0.74 for 8 out of 10 organs). The T2* values revealed a strong relationship with GA in the lungs, liver, and kidney parenchyma (R^2 >0.5). This pipeline was used successfully for a wide range of GAs (17-40 weeks), and is robust to motion artefacts. Low field fetal MRI can be used to perform advanced MRI analysis, and is a viable option for clinical scanning.
Link to paper
DOI: https://doi.org/10.1007/978-3-031-43990-2_34
SharedIt: https://rdcu.be/dnwLP
Link to the code repository
https://github.com/SVRTK/Fetal-T2star-Recon
Link to the dataset(s)
N/A
Reviews
Review #1
- Please describe the contribution of the paper
This study introduces a semi-automatic pipeline using quantitative MRI for fetal body analysis at low field strengths. The pipeline involves acquiring multi-echo dynamic sequences, reconstructing them into a single high-resolution volume, and using a semi-supervised neural network for automatic segmentation of the fetal body into ten organs. The T2* values show a strong relationship with gestational age (GA) in lungs, liver, and kidney parenchyma. The pipeline works successfully for GAs between 17-40 weeks and is robust to motion artefacts, making low field fetal MRI a viable option for clinical scanning.
- 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.
- relevant application with potentially substantial impact in the context of early diagnosis of fetal develpmental abnormalities
- wide applicability since low field MRI is applied
- 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.
- low number of test cases used
- robustness of the pipeline robustness was not properly evaluated
- some findings related to the growth curves are contradictive vs. literature a
- application of the growth curves not shown, although pathologic cases were acquired
- 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 present application-oriented study has poor reproducibility based since:
- data is not made available
- code of the whole pipeline is not made to , only certain parts like the standard nnUnet
- trained models are not made available
- computational requirements were not described
- due to small number of test cases, reproducing the results on different dataset could be challenging (especially since the test data structure is not fully reported, i.e. sex, gestational age, diagnosis, ethnicity, etc.)
- 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
- since the authors had 9 pathological cases, it would be interesting to see whether these could be identified wrt. the normal growth curves
- 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?
- low number of test cases used
- robustness of the pipeline robustness was not properly evaluated
- some findings related to the growth curves are contradictive vs. literature a
- application of the growth curves not shown, although pathologic cases were acquired
- 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 #2
- Please describe the contribution of the paper
This paper presents a semi-automatic pipeline that reconstruct, segment and measures T2* values for major fetal organs using quantitative MRI (T2* relaxometry) at low field strength. A single high-resolution volume (both for structural and quantitative T2*) was reconstructed from multi-echo dynamic sequences of the fetal body using a deformable slice-to-volume reconstruction method. A neural network trained using a semi-supervised approach was used for automatic segmentation of fetal organs. The method would be helpful to reduce manual intervention for low field quantitative MR applications for fetal organ imaging.
- 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 presents an automated pipeline for quantitative mean T2* fetal body organs at low field MRI. This will be helpful to streamline the imaging workflow outside of the clinicians office and therefore, reducing manual labor and effort.
- The paper generate normative T2* growth curves of different fetal body organs at low field MRI, and therefore will be helpful in future studies for distinguishing between normal and pathologic fetal organs. This will also provide basis for understanding underlying pathophysiological process for fetal growth.
- 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 does not provide enough detail of different components of the pipeline. For example, the authors have mentioned deformable slice-to-volume reconstruction (dSVR), however, the detail of the used method were lacking. What type of deformable reconstruction was used? How the reconstruction/ registration errors were measured? How much registration error were considered to be acceptable?
- It was not clear how the DSC values were computed: specifically how the ground truth were generated for DSC computation. Were these manual segmentations performed on single slices or in the reconstructed volume? How many clinicians took part in the manual segmentation? Were there any inter-observer variability?
- The authors didn’t show the growth curves for pathological cases. It will be interesting to see how the growth curves in terms of T2* and volume compare between normal and pathological cases.
- 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 has provided enough details for ensuring general 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
- The abstract states that the pipeline is “very robust to motion artefacts”. Please provide quantitative measures to justify this statement.
- Fetal MRI can be used from “approximately 16 weeks of gestation until birth”. Please provide appropriate reference for supporting this statement.
- Section 2.1 mentions range of values for TR, number of slices and number of dynamics. Why a range was used rather than a fixed value for all cases?
- 2.2 (Reconstructions) section: spelling mistake- ‘contract’ in place of ‘contrast’.
- 23 training cases, 4 validation cases, 7 testing cases were used for 3D nnUNet. Is the number too small for optimal training?
- It will be interesting to see how the pathological cases compare with the normal cases in terms of of organ volume and T2* curves.
- 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 proposed pipeline will be impactful for clinical translation of low-field quantitative fetal MRI.
- 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 #3
- Please describe the contribution of the paper
The manuscript presents a deep-learning pipeline for fetal body segmentation from low field MRi T2* images.
- 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.
- Well written and easy to follow paper
- Relevant topic, in line with MICCAI scope
- Relatively new application
- The approach relies on state of the art work and I believe its implementation would be feasible in fairly reasonable amount of time.
- Detailed clinical discussion 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.
There is a number of weaknesses in the paper that in my opinion make the paper not suitable for publication in its current form:
- The novelty of the pipeline is either not clear or not particularly strong. Most of the method section makes explicit reference to state of the art papers.
- I understand that the application field is relatively new, but this does not justify the lack of the survey of the state of the art in the introduction.
- The experimental section of the paper is not very convincing (e.g., what about comparison with other approaches?).
- 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
Information to reproduce the experiments is provided.
- 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
- From the introduction, it is not particularly clear what is the main novelty and innovation provided in the paper. It would be helpful to the reader if the authors could specify more precisely the novel methodological aspects of their work.
- The introduction is a bit repetitive: I suggest shortening it while clearly highlighting the main contributions of the work and the methodological challenges the authors want to address.
- The survey of relevant work in the literature should be provided.
- In Sec. 2, the authors should clearly state what is the methodological innovation they are presenting.
- Quantitative comparisons with other similar approaches in the literature should be provided.
- 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?
Innovative application but limited methodological novelty.
- 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 proposes a semi-automatic pipeline for low-field fetal MRI that can potentially advance prenatal analysis development.
- 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 research field is new and promising.
- The pipeline is comprehensive。
- 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 background should be strengthened.
- Some details of the submodule in the pipeline are missing, which might limit reproducibility.
- 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
- One of my main concerns is to make the background more clear. E.g., the concept of the T2* and T2* value should be detailed. This can enable it to reduce the required expertise for following the papers.
- 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
- One of my main concerns is to make the background more clear. E.g., the concept of the T2* and T2* values should be detailed. This can reduce the expertise needed for following the papers.
- Please report the relevant details of the hyperparameters, e.g., the setup of dSVR; this enables better reproducibility
- There are many holes in the manual segmentations in Fig. 1. This will raise confusion on the quality of the annotations
- 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, this paper has its merit and could be accepted after addressing the issues.
- Reviewer confidence
Somewhat confident
- [Post rebuttal] After reading the author’s rebuttal, state your overall opinion of the paper if it has been changed
5
- [Post rebuttal] Please justify your decision
I have no further comments here. However, I would like to emphasize that the author should revise the whole paper to make it clearer and easier for colleagues who may not be familiar with this area.
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 describes a semi-automatic pipeline for fetal body MRI reconstruction at low field strength that yields a quantitative T2* relaxometry analysis of all major fetal body organs. The paper describes results on several test sets. The reviewers have raised relevant and important issues with the paper. In particular the authors should address:
- Comments/ evaluation on the robustness of the pipeline
- Clarify the main novelty of the proposed method
- Provide key details of the method as indicated.
Author Feedback
Thank you to all the reviewers (R1,R2,R3, R4) who pointed out our work’s major strengths such as it “will be impactful for clinical translation of low field quantitative fetal MRI”, it is a “relevant application with potentially substantial impact”, “new and promising”, and has “wide applicability”, as well as provided constructive feedback and suggestions. We have grouped our responses into 3 categories based on the reviews, specifically emphasizing where details the reviewers asked can already be found in the submission and where we can emphasize these points further, and are confident this response addresses the concerns.
1.Robustness R1 raised concerns regarding the evaluation of the robustness and R2 pointed out that we fail to justify fully that our pipeline is “very robust to motion artefacts” and questioned how we determined if a dSVR reconstruction was considered acceptable.
These elements were already included in the submission as follows: -We performed a visual review of all reconstructions to determine acceptability: only 2/41 cases (GA<20weeks) were excluded after reconstruction due to excessive motion (Section 3.1). 39/41 subjects had a successful reconstruction (95% success rate, excellent inclusion rate in fetal studies. We will soften the statement in the abstract from ‘very robust’ to ‘robust’. -DSC values were included to evaluate the robustness of the segmentation step (Table 1). -The T2* values obtained from the individual dynamics were compared to the final reconstructed T2* values (Table 2); the values were equivalent. We will re-word the second half of Section 3.1 to make the validation experiment more prominent.
- Novelty of Methodology While R1, R2, and R4 commented positively about the novelty (see quotes above), R3 raised concerns regarding the novelty. Our method indeed builds on the work of pre-existing methods. However: -Our development of dSVR to multichannel (reconstructing both the structural image and the T2* maps) is new and novel. -The automatic processing of low-field functional data to yield absolute fetal body organ T* values is novel. Without these dSVR reconstructions, the ability to determine T2* values of tiny organs (ie adrenal glands) is not possible. -Not only is the low field aspect novel, but such a detailed T2* analysis of fetal body organs hasn’t been done at higher (1.5T/3T) field strengths. These points are briefly mentioned in the introduction, but we can adjust the Introduction and Discussion to emphasis these novel aspects prominently.
R1 and R2 requested the 9 pathological cases already included in our study be added to the growth curves: We will add color-coded points to Figure 4.
- Method Details The requested additional details can be easily addressed with small changes: -R2 requested details on the hyperparameters of dSVR: We will add these -R1 requested more metadata on the testing dataset: We will add the GAs and pathologies to the Supplementary Info -R2 requested more details on the ground truth labels: We state in “Segmentation” that “Seven of the initially generated label maps were corrected in detail to create the ground truth”. One experienced clinician labelled complete volumes. Previous fetal studies done by the clinician include manual segmentations with inter-rater variability explored, we can reference this (reference anonymized) -R1 stated that the data, code, and models were not available: We stated (discussion) that “The data (including images, reconstructions, and segmentations) will be made available from the corresponding author upon reasonable academic request”. We can also emphasise that the entire pipeline will be publicly available. -R2 asked why a range of values were used for TR, # slices, # dynamics: All based on the size of the fetus and continuous improvements of the sequence All of the above points result in minor adjustments to the methods section, and greatly improve the description and reproducibility of the paper.
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.
Only one reviewer commented on the rebuttal and maintained his position – weak accept. On reading the authors rebuttal and the reviewers criticisms, I think that the important of problem addressed and the possible applications to other scenarios tip the balance towards acceptance for me.
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 presents a semi-automatic pipeline for fetal body analysis from MR. The paper was borderline, and prior to rebuttal the reviewers had several concerns, including small number of subjects, lack of robustness analysis, growth curve findings in contrast with existing literature, no application of growth curves in pathological cases, lack of detail in exposition, lack of novelty, lack of comparison to other methods, lack of discussion of related work. While the authors address some of the issues, and promise to fix several of the issues in the final version, the paper still comes across as a little premature, and it would likely be more suited for a revision where we could actually see the updated paper. Thus, at this point I do not recommend acceptance.
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 paper proposes a pipeline to segment and 3D reconstruct different fetal organ anatomies from low field MRI scans, with the purpose of measuring fetal growth. The method is tested on a range of gestational ages, including pathological cases.
Strengths:
- Reviewers acknowledge that the proposed method is very relevant to MICCAI, and it has good potential for impact The method design is sound
Weaknesses:
- Some reviewers note the relatively low number of test cases
- Some reviewers note that there is little algorithm novelty, in that this is mostly a pipeline assembled from existing techniques. Reviewers still acknowledge that the problem / pipeline as a whole is new so this is a minor concern.
- There are relevant method/experiment details missing in the manuscript. Authors state that some of them will be fixed in their rebuttal statement.
Overall I find that this is a nice pipeline integration paper, with good motivation and discussion around its clinical application. I find that the key reviewer concerns have mostly been addressed in rebuttal, so I’m leaning to accept this paper. The test sample size may be small for machine learning standards, but I find it an acceptable number for an initial conference study where the focus is on presenting the novel pipeline application.