List of Papers Browse by Subject Areas Author List
Abstract
The diffusion of minimally invasive, endovascular interventions motivates the development of visualization methods for complex vascular networks. We propose a planar representation of blood vessel trees which preserves the properties that are most relevant to catheter navigation: topology, length and curvature. Taking as input a three-dimensional digital angiography, our algorithm produces a faithful two-dimensional map of the patient’s vessels within a few seconds. To this end, we propose optimized implementations of standard morphological filters and a new recursive embedding algorithm that preserves the global orientation of the vascular network. We showcase our method on peroperative images of the brain, pelvic and knee artery networks. On the clinical side, our method simplifies the choice of devices prior to and during the intervention. This lowers the risk of failure during navigation or device deployment, and may help to reduce the gap between expert and common intervention centers. From a research perspective, our method simulates the cadaveric display of artery trees from anatomical dissections. This opens the door to large population studies on the branching patterns and tortuosity of fine human blood vessels. Our code is released under the permissive MIT license as part of the scikit-shapes Python library (https://scikit-shapes.github.io).
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2025/paper/4231_paper.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to the Code Repository
https://github.com/scikit-shapes/scikit-shapes
Link to the Dataset(s)
N/A
BibTex
@InProceedings{HouGui_Untangling_MICCAI2025,
author = { Houry, Guillaume and Boeken, Tom and Allassonnière, Stéphanie and Feydy, Jean},
title = { { Untangling Vascular Trees for Surgery and Interventional Radiology } },
booktitle = {proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025},
year = {2025},
publisher = {Springer Nature Switzerland},
volume = {LNCS 15968},
month = {September},
}
Reviews
Review #1
- Please describe the contribution of the paper
This approach computes two-dimensional vessel maps for given three-dimensional angriograms.
- Please list the major strengths of the paper: you should highlight 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.
None.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
-
It is unclear what is meant by “three-dimensional angriogram”. From the “Contributions” section, it seems that CT images are meant, as “Hounsfield unit” is mentioned. However, the image in Fig. 1 does not look like normal CT images. Nor does it look like MRA or any other common modalities. So it is questionable on what images the proposed method is supposed to be applied.
-
The segmentation method is trivial and outdated by at least 20 years. This method would only work – if it works at all – on images of very high quality.
-
There are no quantitative results, no comparison or ablation study. It is even unclear on what data the results in Fig. 6~8 are produced.
-
- Please rate the clarity and organization of this paper
Poor
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The submission does not provide sufficient information for reproducibility.
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2025/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(1) Strong Reject — must be rejected due to major flaws
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
Method is long outdated and would not work. Quantitative result and comparison with other methods are completely missing.
- Reviewer confidence
Very confident (4)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
N/A
- [Post rebuttal] Please justify your final decision from above.
N/A
Review #2
- Please describe the contribution of the paper
The article presents a method for generating a planar representation of a 3D vascular network extracted from CTA volumes. This representation is specifically designed to preserve the vessels’ lengths and local curvature, particularly near bifurcations, while ensuring that the resulting layout is free of intersections. The goal is to enhance the understanding of complex vascular geometries, which can be especially valuable for preoperative or intraoperative planning in cardiovascular surgery. In addition, the authors provide an efficient and open-source implementation of their pipeline, capable of producing such representations within a few seconds.
- Please list the major strengths of the paper: you should highlight 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 is well written and illustrated
- The topic is highly relevant and addresses a challenging problem for which few methods have been proposed.
- The proposed representation is conceptually interesting and has the potential to support a wide range of research and clinical applications related to vascular networks.
- The authors provide a light and efficient and open-source implementation of their pipeline, which enhances the reproducibility and potential impact of their work.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
-
The article lacks a sufficiently detailed discussion of related work and does not clearly position the proposed method in the context of existing approaches. Although the authors cite a review [4] and several relevant methods [8, 15, 21, 31], the manuscript does not provide a critical comparison that outlines the limitations of these approaches and how the proposed method addresses them. A more thorough analysis, highlighting both the strengths and weaknesses of prior work and clearly articulating the specific contributions and advantages of the proposed approach, should be included in the rebuttal.
-
The article lack comparisons with related work
The article does not include any comparison—quantitative or visual—between the proposed representation and existing methods. Including at least a visual comparison with alternative approaches would significantly strengthen the manuscript by illustrating the advantages or differences of the proposed method. -
The quality of the representation is not evaluated The authors claim that the proposed representation preserves vessel length and local curvature near bifurcations, but this is not supported by any quantitative evaluation. To substantiate these claims, the authors should provide objective metrics—for example, by calculating the percentage of vessels for which length and curvature are accurately preserved.
-
The claim of clinical utility is not supported by evidence The authors suggest that the proposed visualization could assist clinicians in understanding vascular geometry and support treatment planning decisions. However, this claim is not supported by any form of clinical validation. It remains unclear whether the representations were shown to clinicians or whether any feedback was collected to assess their utility in a clinical context.
-
Some parts of the proposed approach are not described with sufficient clarity (see additional comments section)
-
- Please rate the clarity and organization of this paper
Satisfactory
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The authors claimed to release the source code and/or dataset upon acceptance of the submission.
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2025/en/REVIEWER-GUIDELINES.html
-
Vascular segmentation and robustness of the approach The authors use a learning-free pipeline to segment the vascular network and obtain centerlines and radius estimates at each point. As mentioned in the final section of the article, this pipeline appears to work well, likely due to the high quality of the peroperative images used. However, such image quality may not be typical for the broader range of datasets to which this method could be applied. Robust, topologically correct 3D vascular segmentation remains an open challenge, with an entire field of research dedicated to addressing this problem. In my opinion, the article would benefit from removing the segmentation from the pipeline. Rather than detailing this specific approach, which is highly tailored to their application and not easily generalizable, the authors could assume that the input includes precomputed centerlines and radius estimations. This would allow more space to focus on the true contribution of the article: the planar representation.
-
Assumptions regarding input and vascular circles The method assumes that the vascular network has only one input, which does not hold for all vascular systems. For example, the brain’s vascular network has two main inputs (the carotid arteries). Additionally, the method assumes the absence of cycles, which is not true for the brain (e.g. the Circle of Willis). While the authors briefly mention these limitations in the future work, they do not explain how they addressed these challenges in their experiments. It appears that the authors intentionally ignored one carotid artery, but the handling of the Circle of Willis remains unclear. Furthermore, there are multiple vascular networks in the human body that do not present cycles and have only one input. I wonder why the authors chose the brain vascular network, with its dual inputs and circular structure, to demonstrate their approach. The authors should consider applying their method to datasets such as PARSE (lungs) or ASOKA (coronary arteries), which feature simpler vascular networks, to better illustrate the general applicability of their approach.
-
Preservation of curvature around junctions If I understand correctly, the proposed algorithm aims to preserve the curvature of each vessel branch. However, if the curvature leads to vessel intersection, it is discarded, and the vessel v_i is straightened until the next bifurcation, where the process is repeated. This raises a question: when the algorithm reaches the next bifurcation, is the curvature of the next vessel (v_i+1) preserved relative to its parent vessel (v_i) ? Specifically, the angle between the two vessels is not preserved, correct ?
-
All equations should be numbered for easier reference
- Clarification required
- The authors should consider adding a figure that clearly illustrates the overall pipeline, accompanied by an overview section in the text. Each step of the pipeline should be explicitly named and visually represented to improve clarity and comprehension. For example:
- Extraction of centerlines and radius (eventually discarded),
- Point-wise curvature estimation,
- Recursive 2D embedding process,
- Final intersection-free planar representation.
- The organization of the Methods section could be improved. Specifically, the definition of the 2D embedding is currently introduced within the angular parametrization section, whereas it would be more logically placed in the section on the intersection-free planar layout—after the introduction of the signed angle and curvature functions. A clearer progression would be: first, introduce the signed curvature function; second, present the signed angle function; and finally, describe the 2D embedding along with the intersection-free layout algorithm.
- Section “Angular parametrization”.
-
It is unclear whether the operations in this section are performed on graph G or on Tcoarse. Specifically, is v in V, or v in J ?
- Section “Computing signed curvature on the vessel tree”
- The sentence “Due to torsion, we may not always be able to define a globally consistent orientation for the osculating planes of our space curves.” is unclear. The terms “torsion” and “osculating planes” should be defined more precisely, and an illustrative example would help clarify the meaning and implications of this limitation.
- Section “Intersecton-free planar layout”
- The sentence “we cannot satisfy this constraint” is ambiguous. It is not clear which constraint the authors are referring to.
- The authors state that they compute “v bar” in this section, but the equation defining “va bar” is introduced earlier in the Angular Parametrization section. This could lead to confusion and should be reorganized for clarity.
-
The term “maximal slice of each branch” is vague. What exactly do the authors mean by “slice” in this context? A definition or illustrative example would be helpful.
- Section “Force-directed layout refinement”
- The intuition behind the repulsion term Frep should be further explained. How was this formulation derived ?
-
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(4) Weak Accept — could be accepted, dependent on rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
Despite several limitations, the paper addresses an important and underexplored problem. The proposed planar representation of 3D vascular networks is original, well-motivated, and potentially valuable for clinical and research applications. The metho is supported by an efficient, open-source implementation, which enhances its reproducibility. With appropriate revisions and clarifications, the work has the potential to make a meaningful contribution to the field.
- Reviewer confidence
Very confident (4)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Accept
- [Post rebuttal] Please justify your final decision from above.
The authors have satisfactorily addressed most of my comments. However, I would have appreciated the opportunity to review the revised manuscript in order to directly assess the implemented changes.
Assuming the authors incorporate all the modifications outlined in their rebuttal, I believe the paper is suitable for acceptance to MICCAI 2025.
Review #3
- Please describe the contribution of the paper
The paper presents a method for 2D representation of 3D vascular networks with a strong emphasis on preserving curvature in large proximal vessels while ensuring geometric fidelity at vessel junctions.
- Please list the major strengths of the paper: you should highlight 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 clear and well-structured explanation of the proposed method, supported by high-quality visualizations that effectively illustrate both the process and the results. Additionally, the method demonstrates strong clinical relevance by aiding surgical planning and intervention, while also being computationally efficient, producing results within seconds.
- Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.
The paper lacks a visual comparison with existing methods, which would help contextualize the improvements and trade-offs of the proposed approach.
- Please rate the clarity and organization of this paper
Good
- Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.
The authors claimed to release the source code and/or dataset upon acceptance of the submission.
- Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2025/en/REVIEWER-GUIDELINES.html
N/A
- Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.
(5) Accept — should be accepted, independent of rebuttal
- Please justify your recommendation. What were the major factors that led you to your overall score for this paper?
The method is computationally efficient, producing results within seconds, and has strong clinical relevance. The explanations are clear, and the visualizations are particularly well-executed, enhancing the understanding of the proposed approach.
- Reviewer confidence
Confident but not absolutely certain (3)
- [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.
Accept
- [Post rebuttal] Please justify your final decision from above.
Thanks to the authors for the clarifications
Author Feedback
We thank Reviewers 1 and 3 for their insightful feedback and appreciate their recognition that our work addresses a clinically relevant problem with few proposed methods. Due to character limits, we cannot address all of Reviewer 1’s clarification requests in this rebuttal. However, we are grateful for the in-depth review and will incorporate all comments in our revision.
Reviewer 2 was dismissive of our work and did not comment on our main contribution. To clarify:
All our experiments use three-dimensional X-rays with some blood vessels highlighted by a contrast agent. To be specific: we use a Philips Azurion system and follow a CBCT protocol, with contrast injection via a 5 FR catheter at a rate of 5 cc/sec. Fig. 1 was rendered with shadows enabled in Paraview, which may have caused confusion. We will change this to a more standard view as in Fig. 8.
In our submission, we openly acknowledge that our “classical” preprocessing pipeline benefits from high-quality peroperatory images. Extending to routine data via deep learning-based segmentation is a key future avenue, that we highlighted in our conclusion. Nonetheless, we made a non-trivial implementation effort to provide a fast and open-source segmentation baseline that lets clinical researchers try out our method on high-resolution volumetric data locally, on their own laptops. This was appreciated by Reviewer 3.
To focus our message, we will follow Reviewer 1’s suggestion and replace the first half of page 3 (“Vessel graph extraction”) by a simple reference to the segmentation module of our (anonymous) open source library. This will free up space to address the main concern raised by all three reviewers - the lack of comparisons with related methods and quantitative evaluation:
While we agree that visual comparisons would enhance our work, copyright issues limit our ability to include them. Reference [21] is the only relevant work providing code we could run on our data, explaining our reliance on citing the recent review [4].
Following Reviewer 1’s advice, we will include a critical comparison in our revision, emphasizing that we are the first to preserve curvatures and junction angles in a 2D visualization of a large vessel network. Most concurrent methods scale poorly, which is problematic since our images are acquired at the start of the intervention and must be processed in under 5 minutes. For instance, [31] handles only a dozen branches, [29] takes 12 minutes for 72 branches, and [8, 15] require minutes for a few thousand nodes—whereas our data exceeds 10,000 nodes and 500 branches. Reference [21] preserves network topology but discards vessel lengths and curvatures. Reference [14] is closest to our work but is designed for needle interventions instead of catheters: it also discards junction angles and curvature information.
We appreciate the suggested datasets and will use them in future works. However, MICCAI guidelines prevent us from including new experiments in our revision. Instead, we will quantify the proportion of junction angles accurately preserved (+/-10°) in the article’s examples: 75% to 90% for vessels with radii exceeding 2mm, and 45% to 55% for finer vessels, necessarily compressed in the 2D plane. By construction, we also guarantee an intersection-free layout with 100% preservation of vessel lengths.
Finally, this work was conducted in close collaboration with two interventional radiology departments, one specializing in stroke. Focusing on cerebral networks allowed us to leverage their expertise and tailor our method to their needs. Their injection protocol highlights only one half of the brain arterial network, enabling us to work around the Circle of Willis in this paper. We are aware that dealing with cycles is our main challenge going forward. Instead of brushing the issue under the carpet, we highlighted this question in the last sentence of the paper to attract interest from the MICCAI community.
Meta-Review
Meta-review #1
- Your recommendation
Invite for Rebuttal
- If your recommendation is “Provisional Reject”, then summarize the factors that went into this decision. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. You do not need to provide a justification for a recommendation of “Provisional Accept” or “Invite for Rebuttal”.
N/A
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’
N/A
Meta-review #2
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Reject
- Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’
This work proposes an algorithm for embedding a 3D vascular network in a planar manner for intervention planning. While the algorithm seems well-designed and the qualitative results of the method seem to have clinical relevance, my main concern is that no attempts were made to perform a quantitative evaluation of the quality of the embedding, or to compare with baseline approaches. The rebuttal explains that other methods either take too long or do not show the desired behaviour. However, to make such claims, one would expect from a MICCAI paper to perform according experiments. It may well be the case that e.g. curvature properties are less well retained by a quicker method compared with a slower one. While I agree that beating benchmarks should not be the main goal of quantitative comparisons, and methods can have merit due to other angles to look at a problem as well, not attempting to come up with a quantitative evaluation or at least a comparison of clinical feedback with a traditional planning method is in my opinion a very strong weakness. Thus, I tend to reject this paper.
Meta-review #3
- After you have reviewed the rebuttal and updated reviews, please provide your recommendation based on all reviews and the authors’ rebuttal.
Accept
- Please justify your recommendation. You may optionally write justifications for ‘accepts’, but are expected to write a justification for ‘rejects’
N/A