List of Papers By topics Author List
| Paper Info | Reviews | Meta-review | Author Feedback | Post-Rebuttal Meta-reviews |
Authors
Leonhard Rist, Oliver Taubmann, Hendrik Ditt, Michael Sühling, Andreas Maier
Abstract
Comprehensive, contiguous visualizations of the main cerebral arteries and the surrounding parenchyma offer considerable potential for improving diagnostic workflows in cerebrovascular disease, e.g., for fast assessment of vascular topology and lumen in stroke patients.
Unfolding the brain vasculature into a 2D overview is, however, infeasible using common Curved Planar Reformation (CPR) due to the circular structure of the Circle of Willis (CoW) and the spatial configuration of the vessels typically rendering them unsuitable for mapping onto simple geometric primitives.
We propose CeVasMap, extending the As-Rigid-As-Possible (ARAP) deformation by a smart initialization of the required mesh to map the CoW as well as a merging of neighboring vessels depending on the resulting degree of distortion.
Otherwise, vessels are unfolded and attached individually, creating a textbook-style overview image.
We provide an extensive distortion analysis, comparing the vector fields of individual and merged unfoldings of each vessel to their CPR results.
In addition to enabling unfolding of circular structures, our method is
on par in terms of incurred distortions to optimally oriented CPRs for individual vessels and comparable to unfavorable CPR orientations when merging the complete CoW with a median distortion of 65 µm/mm.
Link to paper
DOI: https://doi.org/10.1007/978-3-031-43904-9_71
SharedIt: https://rdcu.be/dnwIj
Link to the code repository
N/A
Link to the dataset(s)
N/A
Reviews
Review #4
- Please describe the contribution of the paper
Propose a method for comprehensive, contiguous unfolding of the cerebrovascular system that help better visualizes circular structures. Quantitative and qualitative results better than the widely used CPR 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.
Superior quantitative and qualitative performance comparison with baseline (CPR) methods. The display of cerebrovascular unfolding is impressive.
- 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.
Some details need more clarifications and explanations It is unknown whether the algorithm will be made public
- 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
Code and data are not available from the manuscript. Sharing the code for this method is greatly encouraged.
- 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
It is understandable that arteries perpendicular to the CoW plane are hard to be merged. However, when the merge failed, how to decide where to put the artery in the cerebrovascular map? From the visualization results shown in Fig 4 and supplement materials, they are disconnected from the CoW, which is different from the text-book scheme shown in Fig 2 (right). The disconnections might cause confusions. Although the dataset comes from stroke patients, the arteries shown in the manuscript are mostly normal. It is necessary to display vascular rendering on some challenging arteries when arteries are stenotic/tortuous/having parallel arteries nearby/with aneurysm. It is unclear how detailed centerlines (distance between points) are needed to generate the cerebrovascular map. Centerlines are generated manually? Computation time needs to be reported considering the complexity in image processing. Considering the work is built based on ARAP algorithm, more introduction to ARAP might be helpful to understand the proposed method
- 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?
Cerebral arteries are known for their complexity in structures. With this method, the authors presented impressive visualization results for complete CoW arteries, which can be a great improvement towards existing CPR methods. The manuscript is mostly well written, although more clarifications are needed for some issues. I would also recommend more applications on wider range of data.
- 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 #2
- Please describe the contribution of the paper
This paper describes a method to unfold brain vasculature from 3D data onto a 2D “panoramic” overview image by mapping the circle of willis and merging neighbouring vessels.
- 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.
I think the application seems interesting and of relevance to clinical imaging and intervention. Overall the method seems sound and the authors provide quantitative comparisons with a standard method (CPR). The description of the methods is fairly 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 paper is very technical and it is hard to follow for readers that are not experts in this specific domain. In particular the rationale and overall idea (relating to Fig.1) would need more (or clearer) motivation and explanation.
The quantitative results in Fig.3 are not well explained (at least in my view). Are the median values reported the ones from the distributions shown (at least for the blue curves that doesn’t seem to make sense).
- 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
The authors describe the method, but they don’t seem to provide any code. In that sense it is hard for me to judge if the paper can be easily reproduced.
- 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
- Overall this is good work
- I would suggest to provide a gentler introduction to the topic and more explanations or motivations
- Please clarify the median / distribution mismatch (if it is one) for Fig. 3
- What is the clinical impact of your improvements? Did you test that or plan to do that?
- 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 I felt that the importance or impact of the work did not become entirely clear in the paper. The motivation and idea needs to become much clearer. Overall the problem and the application is interesting. Since I am not an expert in this particular field, I would give it the benefit of doubt.
- Reviewer confidence
Not 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 proposes a parametric unfolding of neurovascular imaging to flatten vessels branching and joining at the Circle of Willis, aiming at a topologically and geometrically preserving 3D-to-2D mapping for intuitive visualisation of complex convoluted structures.
- 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.
- Novel idea, valid and valuable concept
- Initial solution to a complex problem, the usefulness and potential go beyond clinical applications
- Clear, effective, and structured delivery and organisation
- Multidisciplinary framework integrating computer graphics, medical imaging, topology, geometry, mapping transforms and optimisation
- Neat, concise, and apparently sound methodological formulation
- Convincing experiments with results reporting deviations, fail cases, and associated interpretations/discussions
- Promising developments and actual potential for industrial impact
- 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.
- Literature review can be improved
- More advanced formulations may be required to tackle more complex scenarios
- Pathology seems to be limited to ischaemic stoke
- 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
Technical details are sufficiently provided for reproducibility. An available implementation would be ideal as an open-source repository.
- 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 paper is interesting and well written. The introduced concepts are explained in a clear and effective way, even without background knowledge in computer graphics.
Methodological formulations seem to be sound, adopting an adequate notation.
The idea of unfolding complex structures with a 3D-to-2D mapping is crucial for an intuitive understanding of the neurovascular network as a whole at a glance, without requiring overlapping maximal intensity projections from different directions, or specific curvilinear planar reformations for individual vessels. The As-Rigid-As-Possible transformation allows to introduce minimal geometrical distortions to the structures of interests, while keeping the topological connectivity as unaltered as possible. The problem is intrinsically complex and has no trivial solution, considering healthy variants of the Circle of Willis and associated pathology. Critical cases of sharp incidental angles and busy junctions with out-of-plane joining vessels are of particular interest, and the authors try to address the problem in a unified manner.
The assembly strategy may benefit from more advanced formulations with seamless and normalised transitions for stitching neighbouring unfolded territorial areas, perhaps maximising the distortion in the parenchyma, while minimising deformations of the vascular structures.
Experiments show promising results for such approach, and representative comparisons are performed with available unfolding methods for vascular visualisation. Interpretation and discussion are consistent with the results. Limitations are accordingly addressed.
Since the study comprise an interdisciplinary framework, it would be nice to mention relevant literature that might have inspired the unfolding. Similarities with conformal maps [1,2] for 3D texturing of digital design in computer graphics are striking.
Refeferences: 1. B. Lévy, et al., “Least squares conformal maps for automatic texture atlas generation.” ACM Transactions on Graphics, 2002. 2. L. Liu, et al., “A local/global approach to mesh parameterization.” Computer Graphics Forum, 27:5, 2008.
Also, it would be indeed exciting to explore potential future developments considering advanced complex warping and twisting with conformal maps (e.g. with a bit of “imagination” and perhaps intermediary folding stages for the piece-wise assembly of out-of-plane, non-manifold splits for sharp-angle or perpendicular incident branches at busy junctions), along the lines of (Youtube) https://www.youtube.com/watch?v=oNd45VqtOjM
Minor comments on the clinical nomenclature of neurovascular structures: “Media” -> Middle Cerebral Artery (MCA) “Basil” -> Basilar Artery (BA) “Vert” -> Vertebral Artery (VA) “ICA” -> define it as Internal Carotid Artery
Other minor comments: The layout of the paper may separate Methods from Experiments (currently they are merged together), where Experiments could contain: Data, Measuring Distortions and Evaluation subsections.
- 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 reviewer is excited to see this paper at MICCAI and supports the innovation of such representation. It looks promising and has actual potential beyond visualisation, as briefly mentioned by the authors toward the end of the manuscript. It would be great to introduce more synergies among multidisciplinary fields of expertise, towards a much-needed simplified representation of highly complex structures as the neurovascular system. Minor corrections should be feasible, although the contribution is already acceptable “as is”.
- 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
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 is about unfolding brain vasculature from a 3D structure to a flattened 2D vessel representation, which can preserve the topology and geometry of blood vessels. This work is interesting and the method is developed based on a multi-disciplinary framework. The authors are suggested to revise the paper with reference to the comments raised by the reviewers.
Author Feedback
We would like to thank the reviewers for their thorough feedback, helpful suggestions and kind words. This letter responds to the comments by grouping them thematically (marking reviewers with #R2-4).
Technical clarifications First, we address #R2’s important comment to clarify the reported median values in Fig. 3. The violin plots visualize the distribution of the evaluation metric D (which is the mean of the local deformations |d_uv| for all pixels) over all patients. The reported median value is the median of the local deformation |d_uv| averaged over all patients to also highlight the deformation-free ratio. Hence it is not the median from the shown distribution of D. This motivation will be added in 2.6. Further, we are thankful for #R4’s comment concerning the centerline sampling as this is a strength of our method. Centerlines are generated automatically, using [12] as mentioned in 2.1, with the distances being the distance between voxel centers. Smaller sampling distances hardly increase the deformation since the mesh already continuously intersects the vessel. A sparser (clinically reasonable) sampling would also barely change the displayed curvature as long as important vessel points are sampled. We will add this briefly to the discussion. We would also like to thank #R4 for commenting on the placement of disjunct vessels, e.g., ICA, in case of individual attachment. Since bifurcation locations are known, image patches can be placed as close to those points on the CoW as desired (by keeping a defined distance between centerlines) while rotating the individual vessel to the desired template vessel direction. The corresponding areas can be highlighted in color to raise awareness of this image assembly. It would be possible to overlap the bifurcation locations to mimic a fully unfolded circle without disruptions. However, undesired image properties such as overlaps between vessels due to curvature (e.g., ICA S-curve) arise. Hence, we prefer and highlight the clear separation. Dotted centerlines in the separately unfolded patches show their vascular connection, see Fig. 1 at the PCA bifurcation. Regarding the assembling strategy, we agree with #R3 that a seamless stitching is desirable, which is part of future investigations. However, creating such normalized transitions can come with folds in the image or strong distortions, rendering the parenchyma display useless. We are thankful for the suggestion of defining properties for vessel and parenchyma and will investigate how to avoid additional heuristics and problems at perpendicular bifurcation areas.
Clinical comments Next, we would like to thank the reviewers for their comments regarding the clinical impact (#R2) and the stroke-related data (#R3/#R4). The clinical impact which is mentioned in the last paragraph will be discussed and tested with clinical experts to improve its value for certain applications. We use stroke data since it is the main reason for (pathologic) topological impairments of the CoW, requiring algorithmic adaptations. Since the lumen is displayed, diseases such as stenosis are generally visible, however their quality should be investigated clinically. The structure of aneurysms cannot be displayed in a satisfactory manner within a 2D overview image (rather using [8]). However, found candidate locations can be marked in the overview for further assessment. These points will be addressed in 4 while also following #R3’s important advice on the vessel nomenclature.
General We would like to thank #R4 for the comment regarding the computation time. The individual parts of this algorithm can be parallelized, leading to lower computation times compared to the segmentation and labeling. We will provide basic statistics on that in the final version. Last, following #R2’s helpful comment, we will improve the transition from the medical introduction to the technical overview and incorporate the excellent multidisciplinary literature suggestions by #R3.