pith. machine review for the scientific record. sign in

arxiv: 2605.00538 · v1 · submitted 2026-05-01 · 💻 cs.CV · cs.LG

Recognition: unknown

Vesselpose: Vessel Graph Reconstruction from Learned Voxel-wise Direction Vectors in 3D Vascular Images

Rajalakshmi Palaniappan , Christoph Karg , Nemesio Navarro-Arambula , Peter Hirsch , Kristin Kraeker , Lisa Mais , Dagmar Kainmueller
Authors on Pith no claims yet

Pith reviewed 2026-05-09 19:27 UTC · model grok-4.3

classification 💻 cs.CV cs.LG
keywords vessel segmentationgraph reconstructiondirection vectorsTEASAR algorithmtopological accuracyvascular imaging3D medical imagesmicro-CT
0
0 comments X

The pith

Predicting voxel-wise direction vectors jointly with segmentation enables topologically accurate vascular graph reconstruction from 3D images.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper aims to show that learning both vessel segmentation and direction vectors at each voxel, followed by a direction-guided version of the TEASAR tracing algorithm, produces vascular graphs with fewer topological errors. This matters because existing methods often merge close vessels or split trees incorrectly, which can mislead medical analyses of blood flow or organ structure. By addressing topology directly through orientation predictions, the approach handles cases where vessels are closely apposed or multiple trees are present in one scan. It demonstrates this on synthetic and real datasets including rat heart micro-CT scans. The authors also introduce false-split and false-merge metrics to quantify these topological issues.

Core claim

Our approach first predicts voxel-wise vessel direction vectors joint with standard vessel segmentation masks. Second, to extract the vascular graph from these predictions, we introduce a direction-vector-guided extension of the TEASAR algorithm. Our approach achieves state-of-the-art performance on three benchmark datasets, spanning both synthetic and real imagery. We further demonstrate the applicability of our approach to challenging 3D micro-CT scans of rat heart vasculature. Finally, we propose meaningful and interpretable measures of topological error, namely false splits and false merges for graphs. Overall, our approach substantially improves the topological accuracy of reconstructed

What carries the argument

The direction-vector-guided extension of the TEASAR algorithm, which uses predicted voxel-wise vessel orientations to trace and connect vessel paths while avoiding incorrect merges or breaks.

Load-bearing premise

That accurate voxel-wise direction vector predictions can be learned jointly with segmentation and that the direction-guided TEASAR extension will reliably produce topologically correct graphs without introducing new failure modes not captured by the proposed false-split and false-merge metrics.

What would settle it

A test volume where direction vectors are predicted accurately yet the extracted graph still shows elevated false-merge rates between separate vessels or false-split rates within single vessels.

Figures

Figures reproduced from arXiv: 2605.00538 by Christoph Karg, Dagmar Kainmueller, Kristin Kraeker, Lisa Mais, Nemesio Navarro-Arambula, Peter Hirsch, Rajalakshmi Palaniappan.

Figure 1
Figure 1. Figure 1: Segmentation-and-skeletonization vs. Vesselpose. Traditional segment￾and-skeletonize pipelines often produce incorrect skeletons, especially when dis￾tinct vessels lie in close proximity. In contrast, Vesselpose leverages voxel-wise direction vectors to robustly reconstruct vascular trees, naturally handling closely apposed branches as well as multiple distinct trees. lengths, and branching patterns can be… view at source ↗
Figure 2
Figure 2. Figure 2: Blood vessel reconstruction and evaluation. (a) A U-Net predicts vessel foreground and voxel-wise direction vectors from the raw image. (b) A modi￾fied TEASAR algorithm extracts a skeleton graph. (c) Predicted skeletons are evaluated against ground-truth using hierarchical graph matching as assignment strategy, which yields topologically meaningful error metrics. The code for the model and evaluation, alon… view at source ↗
Figure 3
Figure 3. Figure 3: Addressing topological errors with the modified TEASAR algorithm. (a) Predicted foreground mask with distinct ground-truth (GT) trees shown in different colors. (b) The algorithm selects one connected component and identifies its roots and endpoints. (c) For each endpoint, paths are traced to all candidate roots, and the optimal path (with the lowest penalty) is chosen. (d) Voxels within a specified radius… view at source ↗
Figure 4
Figure 4. Figure 4: False Merges & False Splits. (a) A ground-truth skeleton next to three pos￾sible predictions. (b) The predicted graph has one FN and one FP edge. The FP is a false merge since it connects two nodes which are not ancestor of each other. Consequently, the FN is a false split. (c) The FP edge is not a false merge since it keeps the ancestor relation w.r.t. to its parent node intact. Consequently, the FN is no… view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative results for the multi-tree synthetic dataset. First row: Seg￾mentation mask and skeletons overlaid, where each color represents a distinct tree. Our approach separates most trees, whereas U-Net + TEASAR merge all trees into one component. Second row: Failure cases for our method, including missed small terminal branches (red arrows) and falsely merged trees (red rect￾angle). 12 [PITH_FULL_IMAG… view at source ↗
Figure 6
Figure 6. Figure 6: Direction vectors generation and angular difference penalty used in modified TEASAR. (a) Direction vectors (blue) are generated by first identi￾fying the closest point on the ground-truth graph (dark red) and then stepping a fixed distance toward the root along the graph. Since the step size is constant across all voxel locations, direction vectors near the centerline exhibit smaller magnitudes, while thos… view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative comparison for single-tree synthetic. A 3D rendering of one of the samples. (a) shows the raw image (b) Ground-truth skeletons overlaid on the segmentation mask (c) our predicted skeleton overlaid on the predicted binary segmentation. Our method (c) produces a reconstruction that closely matches the ground-truth (b), capturing fine structures and maintaining topological con￾sistency. C.3. Multi… view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative comparison of PARSE2022 segmentation. First row: A 2D slice from one sample. (a) shows the raw image, while (b) and (c) show the raw image overlaid with the provided ground-truth segmentation and our segmenta￾tion, respectively. Second row: A 3D crop from the same sample. (d) shows the raw image, and (e) and (f) show the raw image overlaid with the ground￾truth and our segmentation, respectivel… view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative results of the micro-CT data. Illustrated is one 3D annotated crop from the raw micro-CT test data together with varying vessel skeleton graphs in red: (a) shows the annotated ground-truth skeleton; (b) shows the results of our proposed method; (c) shows the result of the baseline, which consists of a U-Net for foreground segmentation followed by the original TEASAR algorithm. Overall, our meth… view at source ↗
Figure 10
Figure 10. Figure 10: Sensitivity to vector noise. (a) The proposed method shows strong robust￾ness to vector noise: small to moderate perturbations (noise level ε ≤ 1.0) of the predicted vectors have no noticeable impact on the resulting edge-wise F1 score. (b–d) Visualization of the direction vector field under increasing noise levels ε ∈ {0, 0.5, 2}. For clarity, all vectors are normalized. Dark blue indicates low vector ma… view at source ↗
read the original abstract

Blood vessel segmentation and -tracing are essential tasks in many medical imaging applications. Although numerous methods exist, the prevailing segment-then-fix paradigm is fundamentally limited regarding its suitability for modeling the task of complete and topologically accurate vascular network reconstruction. Here, we propose an approach to extract topologically more accurate vascular graphs from 3D image data, building upon highly successful ideas from the related biomedical tasks of cell segmentation and -tracking. Our approach first predicts voxel-wise vessel direction vectors joint with standard vessel segmentation masks. Second, to extract the vascular graph from these predictions, we introduce a direction-vector-guided extension of the TEASAR algorithm. Our approach achieves state-of-the-art performance on three benchmark datasets, spanning both synthetic and real imagery. We further demonstrate the applicability of our approach to challenging 3D micro-CT scans of rat heart vasculature. Finally, we propose meaningful and interpretable measures of topological error, namely false splits and false merges for graphs. Overall, our approach substantially improves the topological accuracy of reconstructed vascular graphs, being able to separate closely apposed vessel segments and handle multiple vascular trees within a single volume.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces Vesselpose, which jointly predicts 3D vessel segmentation masks and voxel-wise direction vectors, then applies a direction-vector-guided extension of the TEASAR algorithm to reconstruct topologically accurate vascular graphs. It reports state-of-the-art results on three benchmark datasets (synthetic and real), demonstrates applicability to rat heart micro-CT volumes, and introduces false-split and false-merge metrics for evaluating graph topology.

Significance. If the central claims hold, the work would be significant for medical image analysis because it moves beyond the segment-then-fix paradigm by using learned direction fields to resolve closely apposed vessels and multiple trees, directly addressing a common failure mode in vascular graph reconstruction. The proposal of interpretable false-split/false-merge metrics is a clear positive contribution that could be adopted more broadly.

major comments (2)
  1. [§4] §4 (Experiments) and associated tables: the abstract and introduction assert SOTA performance plus substantial topological gains from the direction-guided TEASAR extension, yet no quantitative evaluation of direction-vector accuracy (e.g., mean angular error, especially in low-contrast or apposed-vessel boundary voxels) is reported. The false-split/false-merge metrics evaluate only final graph topology and therefore cannot isolate whether the claimed improvements originate from the novel guidance mechanism or from segmentation quality and post-processing alone.
  2. [§4] §4 and Table 2 (or equivalent result tables): no error bars, standard deviations, or statistical significance tests are provided for the reported metrics across the three benchmark datasets, nor are the exact baseline implementations and hyper-parameter settings for competing methods detailed. This makes it impossible to assess whether the topological improvements are robust or reproducible.
minor comments (2)
  1. [§3] The description of the direction-vector-guided TEASAR extension in §3 would benefit from a pseudocode listing or explicit step-by-step comparison to the original TEASAR to clarify the precise modifications.
  2. [Figures] Figure captions for qualitative results should explicitly state the source dataset and whether the displayed direction field is ground-truth or predicted.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript to strengthen the experimental reporting and analysis.

read point-by-point responses

  1. Referee: [§4] §4 (Experiments) and associated tables: the abstract and introduction assert SOTA performance plus substantial topological gains from the direction-guided TEASAR extension, yet no quantitative evaluation of direction-vector accuracy (e.g., mean angular error, especially in low-contrast or apposed-vessel boundary voxels) is reported. The false-split/false-merge metrics evaluate only final graph topology and therefore cannot isolate whether the claimed improvements originate from the novel guidance mechanism or from segmentation quality and post-processing alone.

    Authors: We agree that a direct quantitative evaluation of the predicted direction vectors (e.g., mean angular error, with emphasis on challenging voxels) would provide valuable additional insight and help isolate the contribution of the guidance mechanism. While the false-split/false-merge metrics are intended to capture the end-task topological accuracy that is the primary focus of the work, they do not explicitly separate the effects of direction guidance from segmentation quality. In the revised manuscript, we will add a dedicated analysis of direction-vector accuracy, reporting mean angular error on the benchmark datasets with breakdowns for low-contrast and apposed-vessel regions. We will also include an ablation comparing the full method against a direction-free TEASAR baseline to demonstrate that the topological gains are attributable to the learned vectors. revision: yes

  2. Referee: [§4] §4 and Table 2 (or equivalent result tables): no error bars, standard deviations, or statistical significance tests are provided for the reported metrics across the three benchmark datasets, nor are the exact baseline implementations and hyper-parameter settings for competing methods detailed. This makes it impossible to assess whether the topological improvements are robust or reproducible.

    Authors: We acknowledge the need for statistical rigor and full reproducibility details. In the revised manuscript, we will report standard deviations (or error bars) for all metrics across the three datasets and include appropriate statistical significance tests (e.g., paired Wilcoxon tests) for the observed improvements. We will also expand the experimental section to provide complete descriptions of the baseline implementations, including the precise hyper-parameter settings and any adaptations used for fair comparison. The source code and trained models will be released publicly upon acceptance to enable independent reproduction. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation uses learned directions + external TEASAR extension on independent benchmarks

full rationale

The paper's chain is image → joint segmentation + direction vector prediction (supervised learning) → direction-guided TEASAR extension → graph output. TEASAR is cited as prior external work; the extension is described as novel but not derived from the current predictions by definition. Topological claims rest on false-split/false-merge metrics evaluated on three external benchmark datasets (synthetic and real) plus micro-CT scans, not on self-referential fitting or renaming. No self-definitional loops, fitted inputs renamed as predictions, or load-bearing self-citations appear in the provided description. The method is self-contained against external data.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; the approach relies on standard supervised learning for direction prediction and an algorithmic extension of existing TEASAR without introducing new postulated quantities.

pith-pipeline@v0.9.0 · 5525 in / 1058 out tokens · 27289 ms · 2026-05-09T19:27:50.676411+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

73 extracted references · 28 canonical work pages

  1. [1]

    2019 , eprint=

    DeepVesselNet: Vessel Segmentation, Centerline Prediction, and Bifurcation Detection in 3-D Angiographic Volumes , author=. 2019 , eprint=

    2019

  2. [2]

    Self-supervised Vessel Enhancement Using Flow-Based Consistencies

    Jena, Rohit and Singla, Sumedha and Batmanghelich, Kayhan. Self-supervised Vessel Enhancement Using Flow-Based Consistencies. Medical Image Computing and Computer Assisted Intervention -- MICCAI 2021. 2021

    2021

  3. [3]

    Delineating trees in noisy 2D images and 3D image-stacks , year=

    González, Germán and Türetken, Engin and Fleuret, Franc¸ois and Fua, Pascal , booktitle=. Delineating trees in noisy 2D images and 3D image-stacks , year=

  4. [4]

    and Bitter, I

    Sato, M. and Bitter, I. and Bender, M.A. and Kaufman, A.E. and Nakajima, M. , booktitle=. TEASAR: tree-structure extraction algorithm for accurate and robust skeletons , year=

  5. [5]

    best fit

    T.C. Lee and R.L. Kashyap and C.N. Chu , abstract =. Building Skeleton Models via 3-D Medial Surface Axis Thinning Algorithms , journal =. 1994 , issn =. doi:https://doi.org/10.1006/cgip.1994.1042 , url =

    work page doi:10.1006/cgip.1994.1042 1994

  6. [6]

    proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024 , year =

    Naeem, Roman and Hagerman, David and Svensson, Lennart and Kahl, Fredrik , title =. proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024 , year =

    2024

  7. [7]

    Trexplorer Super: Topologically Correct Centerline Tree Tracking of Tubular Objects in CT Volumes

    Naeem, Roman and Hagerman, David and Alv \'e n, Jennifer and Svensson, Lennart and Kahl, Fredrik. Trexplorer Super: Topologically Correct Centerline Tree Tracking of Tubular Objects in CT Volumes. Medical Image Computing and Computer Assisted Intervention -- MICCAI 2025. 2025

    2025

  8. [8]

    Medical Imaging with Deep Learning , pages =

    Vesselformer: Towards Complete 3D Vessel Graph Generation from Images , author =. Medical Imaging with Deep Learning , pages =. 2024 , editor =

    2024

  9. [9]

    Cellpose: a genera list algorithm for cellular segmentation

    Stringer, Carsen and Wang, Tim and Michaelos, Michalis and Pachitariu, Marius , title=. Nature Methods , year=. doi:10.1038/s41592-020-01018-x , url=

    work page doi:10.1038/s41592-020-01018-x

  10. [10]

    Nature Methods , year=

    Pachitariu, Marius and Stringer, Carsen , title=. Nature Methods , year=. doi:10.1038/s41592-022-01663-4 , url=

    work page doi:10.1038/s41592-022-01663-4

  11. [11]

    IEEE Computer Graphics and Applications , year=

    Voreen: A Rapid-Prototyping Environment for Ray-Casting-Based Volume Visualizations , author=. IEEE Computer Graphics and Applications , year=

  12. [12]

    Tizabi, Florian Buettner, Evangelia Christodoulou, Ben Glocker, Fabian Isensee, Jens Kleesiek, Michal Kozubek, Mauricio Reyes, Michael A

    Maier-Hein, Lena and Reinke, Annika and Godau, Patrick and Tizabi, Minu D. and Buettner, Florian and Christodoulou, Evangelia and Glocker, Ben and Isensee, Fabian and Kleesiek, Jens and Kozubek, Michal and Reyes, Mauricio and Riegler, Michael A. and Wiesenfarth, Manuel and Kavur, A. Emre and Sudre, Carole H. and Baumgartner, Michael and Eisenmann, Matthia...

    work page doi:10.1038/s41592-023-02151-z 2024

  13. [13]

    U-Net: Convolutional Networks for Biomedical Image Segmentation

    Ronneberger, Olaf and Fischer, Philipp and Brox, Thomas. U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2015. 2015

    2015

  14. [14]

    Plos one , volume=

    CT imaging of a multi-organ vascular fingerprint in rats , author=. Plos one , volume=. 2024 , publisher=

    2024

  15. [15]

    arXiv: Optimization and Control ,year=

    Faster Wasserstein Distance Estimation with the Sinkhorn Divergence ,author=. arXiv: Optimization and Control ,year=

  16. [16]

    POT: Python Optimal Transport , journal =

    R. POT: Python Optimal Transport , journal =. 2021 , volume =

    2021

  17. [17]

    , author=

    Tutorial on Directed Acyclic Graphs. , author=. Journal of clinical epidemiology , year=

  18. [18]

    Neuroinformatics , volume =

    The DIADEM Metric: Comparing Multiple Reconstructions of the Same Neuron , author =. Neuroinformatics , volume =. 2011 , publisher =

    2011

  19. [19]

    Communications of the ACM , month = sep, pages =

    Bentley, Jon Louis , title =. 1975 , issue_date =. doi:10.1145/361002.361007 , journal =

    work page doi:10.1145/361002.361007 1975

  20. [20]

    Medical Image Computing and Computer-Assisted Intervention-MICCAI 2016: 19th International Conference, Athens, Greece, October 17-21, 2016, Proceedings, Part III 19 , pages=

    The minimum cost connected subgraph problem in medical image analysis , author=. Medical Image Computing and Computer-Assisted Intervention-MICCAI 2016: 19th International Conference, Athens, Greece, October 17-21, 2016, Proceedings, Part III 19 , pages=. 2016 , organization=

    2016

  21. [21]

    Simultaneous segmentation and anatomical labeling of the cerebral vasculature , journal =

    David Robben and Engin Türetken and Stefan Sunaert and Vincent Thijs and Guy Wilms and Pascal Fua and Frederik Maes and Paul Suetens , keywords =. Simultaneous segmentation and anatomical labeling of the cerebral vasculature , journal =. 2016 , issn =. doi:https://doi.org/10.1016/j.media.2016.03.006 , url =

    work page doi:10.1016/j.media.2016.03.006 2016

  22. [22]

    Reconstructing Curvilinear Networks Using Path Classifiers and Integer Programming , year=

    Türetken, Engin and Benmansour, Fethallah and Andres, Bjoern and Głowacki, Przemysław and Pfister, Hanspeter and Fua, Pascal , journal=. Reconstructing Curvilinear Networks Using Path Classifiers and Integer Programming , year=

  23. [23]

    Simultaneous Segmentation and Anatomical Labeling of the Cerebral Vasculature

    Robben, David and T \"u retken, Engin and Sunaert, Stefan and Thijs, Vincent and Wilms, Guy and Fua, Pascal and Maes, Frederik and Suetens, Paul. Simultaneous Segmentation and Anatomical Labeling of the Cerebral Vasculature. Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2014. 2014

    2014

  24. [24]

    Automated Reconstruction of Dendritic and Axonal Trees by Global Optimization with Geometric Priors , journal=

    T. Automated Reconstruction of Dendritic and Axonal Trees by Global Optimization with Geometric Priors , journal=. 2011 , month=. doi:10.1007/s12021-011-9122-1 , url=

    work page doi:10.1007/s12021-011-9122-1 2011

  25. [25]

    Reconstructing Geometrically Consistent Tree Structures from Noisy Images

    T \"u retken, Engin and Blum, Christian and Gonz \'a lez, Germ \'a n and Fua, Pascal. Reconstructing Geometrically Consistent Tree Structures from Noisy Images. Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2010. 2010

    2010

  26. [26]

    Machine learning analysis of whole mouse brain vasculature , journal=

    Todorov, Mihail Ivilinov and Paetzold, Johannes Christian and Schoppe, Oliver and Tetteh, Giles and Shit, Suprosanna and Efremov, Velizar and Todorov-V. Machine learning analysis of whole mouse brain vasculature , journal=. 2020 , month=. doi:10.1038/s41592-020-0792-1 , url=

    work page doi:10.1038/s41592-020-0792-1 2020

  27. [27]

    Quantitative neuroanatomy for connectomics in

    Schneider-Mizell, Casey M and Gerhard, Stephan and Longair, Mark and Kazimiers, Tom and Li, Feng and Zwart, Maarten F and Champion, Andrew and Midgley, Frank M and Fetter, Richard D and Saalfeld, Stephan and Cardona, Albert , editor =. Quantitative neuroanatomy for connectomics in. eLife , issn =. doi:10.7554/eLife.12059 , url =

    work page doi:10.7554/elife.12059

  28. [28]

    Bioinformatics , volume =

    Saalfeld, Stephan and Cardona, Albert and Hartenstein, Volker and Tomančák, Pavel , title =. Bioinformatics , volume =. 2009 , month =. doi:10.1093/bioinformatics/btp266 , url =

    work page doi:10.1093/bioinformatics/btp266 2009

  29. [29]

    Scalable robust graph and feature extraction for arbitrary vessel networks in large volumetric datasets , journal=

    Drees, Dominik and Scherzinger, Aaron and H. Scalable robust graph and feature extraction for arbitrary vessel networks in large volumetric datasets , journal=. 2021 , month=. doi:10.1186/s12859-021-04262-w , url=

    work page doi:10.1186/s12859-021-04262-w 2021

  30. [30]

    GERoMe-a Method for Evaluating Stability of Graph Extraction Algorithms Without Ground Truth , year=

    Drees, Dominik and Scherzinger, Aaron and Jiang, Xiaoyi , journal=. GERoMe-a Method for Evaluating Stability of Graph Extraction Algorithms Without Ground Truth , year=

  31. [31]

    Efficient anatomical labeling of pulmonary tree structures via deep point-graph representation-based implicit fields , journal =

    Kangxian Xie and Jiancheng Yang and Donglai Wei and Ziqiao Weng and Pascal Fua , keywords =. Efficient anatomical labeling of pulmonary tree structures via deep point-graph representation-based implicit fields , journal =. 2025 , issn =. doi:https://doi.org/10.1016/j.media.2024.103367 , url =

    work page doi:10.1016/j.media.2024.103367 2025

  32. [32]

    Kuhn , title =

    Harold W. Kuhn , title =. Naval Research Logistics Quarterly , volume =. 1955 , publisher =

    1955

  33. [33]

    Computational Topology: An Introduction , isbn =

    Edelsbrunner, Herbert and Harer, John , year =. Computational Topology: An Introduction , isbn =

  34. [34]

    2020 , eprint=

    Computational Optimal Transport , author=. 2020 , eprint=

    2020

  35. [35]

    Mechanosensitive PIEZO2 channels shape coronary artery development , journal=

    Pampols-Perez, Mireia and F. Mechanosensitive PIEZO2 channels shape coronary artery development , journal=. 2025 , month=. doi:10.1038/s44161-025-00677-3 , url=

    work page doi:10.1038/s44161-025-00677-3 2025

  36. [36]

    and Kinney-Lang, Eli and Rashid, Faisal and Hamer, Mary and DeFazio, Richard A

    Obenaus, Andre and Ng, Michelle and Orantes, Amanda M. and Kinney-Lang, Eli and Rashid, Faisal and Hamer, Mary and DeFazio, Richard A. and Tang, Jiping and Zhang, John H. and Pearce, William J. , title=. Scientific Reports , year=. doi:10.1038/s41598-017-00161-4 , url=

    work page doi:10.1038/s41598-017-00161-4

  37. [37]

    eLife , issn =

    Hemodynamics regulate spatiotemporal artery muscularization in the developing circle of Willis , author =. eLife , issn =. doi:10.7554/eLife.94094 , url =

    work page doi:10.7554/elife.94094

  38. [38]

    eLife , issn =

    Three-dimensional imaging of vascular development in the mouse epididymis , author =. eLife , issn =. doi:10.7554/eLife.82748 , url =

    work page doi:10.7554/elife.82748

  39. [39]

    and Stefan, Sabina and Lee, Jang-Hoon and Puttigampala, Pooja and Kim, Anna H

    Walek, Konrad W. and Stefan, Sabina and Lee, Jang-Hoon and Puttigampala, Pooja and Kim, Anna H. and Park, Seong Wook and Marchand, Paul J. and Lesage, Frederic and Liu, Tao and Huang, Yu-Wen Alvin and Boas, David A. and Moore, Christopher and Lee, Jonghwan , title=. Nature Communications , year=. doi:10.1038/s41467-023-38609-z , url=

    work page doi:10.1038/s41467-023-38609-z

  40. [40]

    Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) , month =

    Wittmann, Bastian and Wattenberg, Yannick and Amiranashvili, Tamaz and Shit, Suprosanna and Menze, Bjoern , title =. Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) , month =. 2025 , pages =

    2025

  41. [41]

    Bumgarner and Randy J

    Jacob R. Bumgarner and Randy J. Nelson , keywords =. Open-source analysis and visualization of segmented vasculature datasets with VesselVio , journal =. 2022 , issn =. doi:https://doi.org/10.1016/j.crmeth.2022.100189 , url =

    work page doi:10.1016/j.crmeth.2022.100189 2022

  42. [42]

    and McGinnis, Julian and Shit, Suprosanna and Ezhov, Ivan and B\"

    Paetzold, Johannes C. and McGinnis, Julian and Shit, Suprosanna and Ezhov, Ivan and B\". Whole Brain Vessel Graphs: A Dataset and Benchmark for Graph Learning and Neuroscience , url =. Proceedings of the Neural Information Processing Systems Track on Datasets and Benchmarks , editor =

  43. [43]

    and Prabhakar, Chinmay and Rueckert, Daniel and Menze, Bjoern , booktitle=

    Wittmann, Bastian and Paetzold, Johannes C. and Prabhakar, Chinmay and Rueckert, Daniel and Menze, Bjoern , booktitle=. Link Prediction for Flow-Driven Spatial Networks , year=

  44. [44]

    Deformable

    Xizhou Zhu and Weijie Su and Lewei Lu and Bin Li and Xiaogang Wang and Jifeng Dai , booktitle=. Deformable. 2021 , url=

    2021

  45. [45]

    End-to-End Object Detection with Transformers

    Carion, Nicolas and Massa, Francisco and Synnaeve, Gabriel and Usunier, Nicolas and Kirillov, Alexander and Zagoruyko, Sergey. End-to-End Object Detection with Transformers. Computer Vision -- ECCV 2020. 2020

    2020

  46. [46]

    Scientific Data , year=

    Lyu, Xingzheng and Cheng, Li and Zhang, Sanyuan , title=. Scientific Data , year=. doi:10.1038/s41597-022-01507-y , url=

    work page doi:10.1038/s41597-022-01507-y

  47. [47]

    and Sekuboyina, Anjany and Ezhov, Ivan and Unger, Alexander and Zhylka, Andrey and Pluim, Josien P

    Shit, Suprosanna and Paetzold, Johannes C. and Sekuboyina, Anjany and Ezhov, Ivan and Unger, Alexander and Zhylka, Andrey and Pluim, Josien P. W. and Bauer, Ulrich and Menze, Bjoern H. , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2021 , pages =

    2021

  48. [48]

    Lawrence

    Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll \'a r, Piotr and Zitnick, C. Lawrence. Microsoft COCO: Common Objects in Context. Computer Vision -- ECCV 2014. 2014

    2014

  49. [49]

    Scientific Reports , year=

    Foucart, Adrien and Debeir, Olivier and Decaestecker, Christine , title=. Scientific Reports , year=. doi:10.1038/s41598-023-35605-7 , url=

    work page doi:10.1038/s41598-023-35605-7

  50. [50]

    Similarity evaluation of retinal vascular network based on tree edit distance , year=

    Li, Wenjian and Wu, Guannan and Li, Huiqi , booktitle=. Similarity evaluation of retinal vascular network based on tree edit distance , year=

  51. [51]

    PLOS ONE , publisher =

    Cell Tracking Accuracy Measurement Based on Comparison of Acyclic Oriented Graphs , year =. PLOS ONE , publisher =. doi:10.1371/journal.pone.0144959 , author =

    work page doi:10.1371/journal.pone.0144959

  52. [52]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Mais, Lisa and Hirsch, Peter and Managan, Claire and Kandarpa, Ramya and Rumberger, Josef Lorenz and Reinke, Annika and Maier-Hein, Lena and Ihrke, Gudrun and Kainmueller, Dagmar , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =. 2024 , pages =

    2024

  53. [53]

    Medical image analysis , volume=

    Tissue metabolism driven arterial tree generation , author=. Medical image analysis , volume=. 2012 , publisher=

    2012

  54. [54]

    Numerische mathematik , volume=

    A note on two problems in connexion with graphs , author=. Numerische mathematik , volume=. 1959 , publisher=

    1959

  55. [55]

    Verified details , author=

  56. [56]

    Vessel-GAN: Angiographic reconstructions from myocardial CT perfusion with explainable generative adversarial networks , journal =

    Chulin Wu and Heye Zhang and Jiaqi Chen and Zhifan Gao and Pengfei Zhang and Khan Muhammad and Javier. Vessel-GAN: Angiographic reconstructions from myocardial CT perfusion with explainable generative adversarial networks , journal =. 2022 , issn =. doi:https://doi.org/10.1016/j.future.2021.12.007 , url =

    work page doi:10.1016/j.future.2021.12.007 2022

  57. [57]

    Alexander and Li, Peter H

    Silversmith, William and Bae, J. Alexander and Li, Peter H. and Wilson, A.M. , doi =

  58. [58]

    Frontiers in Oncology , volume=

    Assessment of Vascular Network Connectivity of Hepatocellular Carcinoma Using Graph-Based Approach , author=. Frontiers in Oncology , volume=

  59. [59]

    Hypertension , volume=

    Abstract P203: Microvascular Imaging As Early Predictor For Cardiac Dysfunction After Preeclampsia , author=. Hypertension , volume=. 2023 , publisher=

    2023

  60. [60]

    and Kainmueller, Dagmar and Keller, Philipp J

    Malin-Mayor, Caroline and Hirsch, Peter and Guignard, Leo and McDole, Katie and Wan, Yinan and Lemon, William C. and Kainmueller, Dagmar and Keller, Philipp J. and Preibisch, Stephan and Funke, Jan , title=. Nature Biotechnology , year=. doi:10.1038/s41587-022-01427-7 , url=

    work page doi:10.1038/s41587-022-01427-7

  61. [61]

    PatchPerPix for Instance Segmentation

    Mais, Lisa and Hirsch, Peter and Kainmueller, Dagmar. PatchPerPix for Instance Segmentation. Computer Vision -- ECCV 2020. 2020

    2020

  62. [62]

    The Thirteenth International Conference on Learning Representations , year=

    Topograph: An Efficient Graph-Based Framework for Strictly Topology Preserving Image Segmentation , author=. The Thirteenth International Conference on Learning Representations , year=

  63. [63]

    and Roy, Saikat and Kovacs, Balint and Ulrich, Constantin and Wald, Tassilo and Zenk, Maximilian and Vollmuth, Philipp and Kleesiek, Jens and Isensee, Fabian and Maier-Hein, Klaus

    Kirchhoff, Yannick and Rokuss, Maximilian R. and Roy, Saikat and Kovacs, Balint and Ulrich, Constantin and Wald, Tassilo and Zenk, Maximilian and Vollmuth, Philipp and Kleesiek, Jens and Isensee, Fabian and Maier-Hein, Klaus. Skeleton Recall Loss for Connectivity Conserving and Resource Efficient Segmentation of Thin Tubular Structures. Computer Vision --...

    2024

  64. [64]

    and Zhou, Jiayan and Fan, Xiaochen and Zanetti, Daniela and Naftaly, Jeffrey A

    Rios Coronado, Pamela E. and Zhou, Jiayan and Fan, Xiaochen and Zanetti, Daniela and Naftaly, Jeffrey A. and Prabala, Pratima and Mart. <em>CXCL12</em> drives natural variation in coronary artery anatomy across diverse populations , journal=. 2025 , month=. doi:10.1016/j.cell.2025.02.005 , url=

    work page doi:10.1016/j.cell.2025.02.005 2025

  65. [65]

    and Deb, Diptodip and Lee, Wei-Chung Allen and Saalfeld, Stephan and Turaga, Srinivas C

    Sheridan, Arlo and Nguyen, Tri M. and Deb, Diptodip and Lee, Wei-Chung Allen and Saalfeld, Stephan and Turaga, Srinivas C. and Manor, Uri and Funke, Jan , title=. Nature Methods , year=. doi:10.1038/s41592-022-01711-z , url=

    work page doi:10.1038/s41592-022-01711-z

  66. [66]

    Manton and Aaron D

    Marta Costa and James D. Manton and Aaron D. Ostrovsky and Steffen Prohaska and Gregory S.X.E. Jefferis , keywords =. NBLAST: Rapid, Sensitive Comparison of Neuronal Structure and Construction of Neuron Family Databases , journal =. 2016 , issn =. doi:https://doi.org/10.1016/j.neuron.2016.06.012 , url =

    work page doi:10.1016/j.neuron.2016.06.012 2016

  67. [67]

    Sexton and Dominic Rütsche and Jessica E

    Zachary A. Sexton and Dominic Rütsche and Jessica E. Herrmann and Andrew R. Hudson and Soham Sinha and Jianyi Du and Daniel J. Shiwarski and Anastasiia Masaltseva and Fredrik Samdal Solberg and Jonathan Pham and Jason M. Szafron and Sean M. Wu and Adam W. Feinberg and Mark A. Skylar-Scott and Alison L. Marsden , title =. Science , volume =. 2025 , doi =. ...

    work page doi:10.1126/science.adj6152 2025

  68. [68]

    2024 , eprint=

    Efficient automatic segmentation for multi-level pulmonary arteries: The PARSE challenge , author=. 2024 , eprint=

    2024

  69. [69]

    2017 , isbn =

    Reinhard Diestel , title =. 2017 , isbn =

    2017

  70. [70]

    Machine Learning , author =

    Breiman, Leo , title=. Machine Learning , year=. doi:10.1023/A:1010933404324 , url=

    work page doi:10.1023/a:1010933404324

  71. [71]

    and Vincken, Koen and Viergever, Max , year =

    Frangi, Alejandro and Niessen, W.J. and Vincken, Koen and Viergever, Max , year =. Multiscale Vessel Enhancement Filtering , volume =

  72. [72]

    International Conference on Medical Imaging with Deep Learning,

    Hirsch, Peter and Kainmueller, Dagmar , title =. International Conference on Medical Imaging with Deep Learning,

  73. [73]

    Nature methods , volume=

    nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation , author=. Nature methods , volume=. 2021 , publisher=

    2021