List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Stefan Ploner, Siyu Chen, Jungeun Won, Lennart Husvogt, Katharina Breininger, Julia Schottenhamml, James Fujimoto, Andreas Maier
Abstract
Optical coherence tomography (OCT) is a micrometer-scale, volumetric imaging modality that has become a clinical standard in ophthalmology. OCT instruments image by raster-scanning a focused light spot across the retina, acquiring sequential cross-sectional images to generate volumetric data. Patient eye motion during the acquisition poses unique challenges: Non-rigid, discontinuous distortions can occur, leading to gaps in data and distorted topographic measurements. We present a new distortion model and a corresponding fully-automatic, reference-free optimization strategy for computational motion correction in orthogonally raster-scanned, retinal OCT volumes. Using a novel, domain-specific spatiotemporal parametrization of forward-warping displacements, eye motion can be corrected continuously for the first time. Parameter estimation with temporal regularization improves robustness and accuracy over previous spatial approaches. We correct each A-scan individually in 3D in a single mapping, including repeated acquisitions used in OCT angiography protocols. Specialized 3D forward image warping reduces median runtime to < 9 s, fast enough for clinical use. We present a quantitative evaluation on 18 subjects with ocular pathology and demonstrate accurate correction during microsaccades. Transverse correction is limited only by ocular tremor, whereas submicron repeatability is achieved axially (0.51 µm median of medians), representing a dramatic improvement over previous work. This allows assessing longitudinal changes in focal retinal pathologies as a marker of disease progression or treatment response, and promises to enable multiple new capabilities such as supersampled/super-resolution volume reconstruction and analysis of pathological eye motion occuring in neurological diseases.
Link to paper
DOI: https://link.springer.com/chapter/10.1007/978-3-031-16434-7_50
SharedIt: https://rdcu.be/cVRsi
Link to the code repository
N/A
Link to the dataset(s)
N/A
Reviews
Review #1
- Please describe the contribution of the paper
The paper introduces a novel model and an optimization for volumetric distortion correction in OCT. The method allows for performing. It estimates A-scan positions in a single mapping, combined with a forward warping that allows correction in all 3 dimensions while avoiding the need for repeat scans. It also performs parameterization with respect to time, which allows for fully continuous parameters. Finally, by structuring the feasible distortions in OCT, a forward interpolation scheme is utilized, providing more accurate results while requiring less computational time than competing methods
- 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 provides a good introduction, and a good explanation of the differences between the proposed method and the competing ones.
- The proposed method requires a different protocol for scanning, with a single mapping, which takes less time, so it potentially leads to less errors caused by saccadic movement or by wandering eye from the patients, as well as less discomfort.
- The proposed method is 5 times faster than the fastest method it was compared against
- The proposed method supposedly provides a dramatic improvement over competing methods on registration accuracy.
- 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 were no accuracy comparisons with competing methods, just time comparisons. Although estimations are provided according to the characteristics of the diverse methods, the lack of hard numbers decreases the confidence on the provided results, and it does not really facilitate obtaining an objective comparison by the reader. While this is justified and explained by the authors, one wonders if one or some of the competing methods could have been replicated, or the authors contacted to run experiments in the utilized dataset.
- The dataset is not available and there is no word on if it will be made available in the future. While its characteristics are described in good detail, having a common dataset with ground truth metrics would facilitate that future methods can be compared with the presented one.
- Please rate the clarity and organization of this paper
Excellent
- 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
- Nor the code, nor the dataset are publicly available.
- The framework is said to be made available in the future, once further work is performed and the framework is considered closer to completion
- The mathematical support of the method is clear and detailed enough to try to attempt an implementation somewhat close to the one the authors might have done.
- While the dataset is not publicly available, it is sufficiently well described to be able to acquire a similar one, if access to a suitable device is had.
- 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 paper was quite clear and well organized. The introduction section was very detailed and contained good details on what the state of the art is, and what are the differences introduced by the proposed method.
- The method is well defined and described. The contributions are made clearly, and the method provides several improvements in speed, accuracy and in comfort to the patient, as less acquisitions are required.
- The weakest part of the paper would be the experiments section. There were no accuracy comparisons with competing methods, just time comparisons. Although estimations are provided according to the characteristics of the diverse methods, the lack of hard numbers decreases the confidence on the provided results, and it does not really facilitate obtaining an objective comparison by the reader. While this is justified and explained by the authors, one wonders if one or some of the competing methods could have been replicated, or the authors contacted to run experiments in the utilized dataset. The dataset is not available and there is no word on if it will be made available in the future. While its characteristics are described in good detail, having a common dataset with ground truth metrics would facilitate that future methods can be compared with the presented one.
- The discussion is quite interesting. It goes into detail on why there was no direct comparison, and it makes an effort on providing estimates on the range of error of competing methods. It also provides a good analysis on what the future steps for the framework might be.
- 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 give score is due to the proposed method being faster than alternatives, supposedly more accurate, and simper given the need for just a single mapping, which reverts on more simplicity, as well as more comfort for the patient.
However, I find the experiments section not to be as strong as it could have been. With proper experimentation and comparisons, utilizing hard numbers when comparing with alternatives, I would have definitely rated the paper higher.
- Number of papers in your stack
4
- What is the ranking of this paper in your review stack?
4
- 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 #2
- Please describe the contribution of the paper
The authors proposed a motion model and a corresponding fully automatic, reference-free optimization strategy for volumetric distortion correction in OCT.
- 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 authors proposed a novel strategy to correct OCT volumetric distortion.
- The authors made experiments based on patients to prove the effectiveness.
- The authors proposed the first metric quantification of errors in OCT motion correction.
- 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.
- What is the meaning of right part in the equation (1)?
- In the Fig.3, how did authors take the top row two images almost at the same position, as the eye is moving? If the positions are different, how can you get the final OCTA.
- Would you further explain the eye motion trajectories?
- The runtime comparison is shown in Table 1. The authors reproduced the comparison algorithms and gave out the results based on the same GPU?
- The references of Zang17, Makita21 and Chen21 in table 1 are missed.
- How are the results from other diseased patients?
- 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
How to get the X/Y fast motion volumes may determine the reproducibility of 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
There is no conclusion in the paper.
- 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?
The detailed explainations of the paper and the results seems good.
- Number of papers in your stack
1
- What is the ranking of this paper in your review stack?
4
- 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
5
- [Post rebuttal] Please justify your decision
Not Answered
Review #3
- Please describe the contribution of the paper
eye emotion during scanning OCT images causes artifacts, This paper, introduce a registration method to compensate 3D-OCT eye motion. The author claim that their result are promising and improved dramatically compare to the state-of-art.
- 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.
using non-rigid registration compare to rigid-registration, using regularization term , using a-scan and B-scan information , motion compensation throughout fast microsaccadic eye movements.
- 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.
missing reference of Table 1
lack of enough comparison with more methods in terms of accuracy
- 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
however the code is not provided , but provided visual inspection is satisfactory
- 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
1- author should provide complete reference which are written in the paper (table1, Zhang? Ploner? Makita? Chen?) 2- in terms of accuracy comparison , some team have worked on this topic for long term which are missing in references such as Martin F. Kraus et al, your results should be comparted to them
- 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?
1-The novelty paper should be more strong
2- the registration parameters are not described competently, for example in literature mentioned about Mutual information or SSD cost function. these should be included too 3- - Number of papers in your stack
5
- What is the ranking of this paper in your review stack?
3
- Reviewer confidence
Very 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.
The paper deals with volumetric distortion correction in OCT. Reviewers have raised a number of concerns, related to problem setup, and description. There seem to be typos in the formulation leading to difficulty for the reviewers. Please see comments by reviewer 2. There are concerns about reproducibility since data and code are not publically available.
- 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).
8
Author Feedback
We want to thank the reviewers for their positive feedback, especially with regard to the accuracy/speed of the proposed method, and its demonstration in patient data. Additionally, we highly appreciate the reviewers’ suggestions, and answer their concerns and questions in our pointwise response below.
META
Clarity: We will clarify all raised uncertainties in the text, make figures/plots more self-explanatory and fix the missing references for a final version of the paper.
Reproducibility: Unfortunately, publication of our clinical patient data is not possible under our current IRB. A public, clinical dataset using orthogonally scanned acquisition does not exist. As literature showed large differences between healthy and pathologic data [11], using data from healthy eyes instead of clinical data is not a representative option.
R3 (weak reject)
Novelty: Since publication of the seminal work by Kraus et al. in 2012/2014 [6], the major advance was correction of torsional eye rotation/head tilt by Lezama et al. [8]. However, their method was incompatible with the Kraus approach, and was a step back in correction of the more prominent fixational eye motion, which made this method noncompetitive in typical use cases.
We identified limitations of the current state of the art (SOTA) and made multiple advances; dramatically improving accuracy, combining correction of fixational eye motion and torsional motion and allowing multiple new applications. The main challenges solved and the novelty in this paper are the formulation of a complete spatiotemporal motion model, and the OCT-specific efficient algorithms to implement the therefore necessary forward warping with runtimes fast enough for clinical use. Despite many related publications in the eight years since Kraus, this has not been demonstrated before. To transport these achievements more clearly, we will better highlight them in the final version of the paper.
Comparison to Kraus: We provide a detailed comparison with the method by Ploner, Kraus et al. [11], which we believe is representative of the SOTA. This method extends the reviewer-mentioned Kraus method [6] with an OCTA similarity term, but uses the previous motion model. Therefore, our model comparison applies to both methods in the exact same way, making an extra comparison with [6] superfluous. We clarified this in the text.
Data term: The SSD is reflected in eq. 3. MI is not mentioned in the paper nor indicated: Single modality registration, the local deviations in OCT illumination are better modeled by SSD, which reflects closeness in intensity, while MI does not.
R1 (accept)
Lack of quantitative comparison: Prior methods provide quantitative results on a proxy task (e.g., layer segmentation). However, layer segmentation itself has an inherent error of ~3 um [4], which is an order of magnitude larger than the errors we measure with our method. Therefore, it does not make sense to measure “our” error using layer segmentation. Also, prior backward warping methods cannot quantify differences in displacement fields because these methods sample in the input space, which has positional ambiguities in case of self-overlap, discontinuities or gaps. Therefore, we have to resort to qualitative comparisons across the methods. These comparisons show that our method achieves clearly superior results. We will ensure that this is clearly explained in the final version.
R2 (weak accept)
Clarity (all were clarified in the paper):
- Eq. 1 maps scan pattern locations (\vec x) with the estimated eye motion (matrix) to (motion-corrected) retinal locations (left side)
- Patients fixate a target, leaving the acquisition location mostly unchanged. The issue are discontinuities (Fig. 1), gaps, and distortion.
- Trajectories display (per B-scan) estimated t^x / t^y / t^z parameters.
- Code for the compared methods is not publically available, reimplementation of these multi-stage methods is out of scope for this work.
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.
I think the authors have addressed the concern raised by the reviewer about novelty. The motion correction on OCT is important problem, and the approach could be applicable to other fields where motion is performed.
- 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).
5
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.
Motion correction in OCT images is a hard problem without much advancements recently. The main downside is the need for orthogonal raster-scan, which does not make it applicable to retrospective datasets. Nevertheless, looking at the rebuttal, the authors successfully addressed the remark of R3 regarding the novelty, and I find the paper to advance the state of the art on this important topic.
- 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).
5
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.
There have not been any comments from Reviewer 3 (weak reject score because there were insufficient comparisons against other methods) since the rebuttal, but the authors seem to have provided a good justification for their choice of baseline (I know little about the OCT registration field). Other concerns seem to have been addressed.
- 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).
12