pith. sign in

arxiv: 2512.03869 · v4 · pith:Q4C4RCYUnew · submitted 2025-12-03 · 💻 cs.CV · cs.CY

An Automated Framework for Large-Scale Graph-Based Cerebrovascular Analysis

Daniele Falcetta , Liane S. Canas , Lorenzo Suppa , Matteo Pentassuglia , Jon Cleary , Marc Modat , S\'ebastien Ourselin , Maria A. Zuluaga This is my paper

Pith reviewed 2026-05-21 18:06 UTC · model grok-4.3

classification 💻 cs.CV cs.CY
keywords cerebrovascular analysisvessel graphautomated skeletonizationTOF-MRAvascular agingnormative modelingpopulation studiesmorphometric features
0
0 comments X

The pith

A skeletonization-to-graph pipeline computes fifteen cerebrovascular features automatically to support population studies of vascular aging.

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

The paper presents CaravelMetrics as a fully automated way to convert 3D time-of-flight magnetic resonance angiograms into vessel graphs and then extract morphometric, topological, fractal, and geometric measurements. These measurements can be taken across the whole network or inside specific arterial territories. The authors apply the method to 570 scans spanning ages 20 to 86 and recover age-related, sex-related, and education-related patterns that match earlier manual studies. A sympathetic reader would care because the approach removes the need for hand-tracing vessels, making it feasible to build normative references and run large-scale investigations into how vascular organization changes with health and aging.

Core claim

The central claim is that integrating atlas-based parcellation, centerline extraction, and graph construction from skeletonized vessels produces reproducible quantitative features that capture biologically relevant variations in cerebrovascular organization across large cohorts.

What carries the argument

Skeletonization-derived graph representations of the vascular network that enable computation of fifteen multiscale morphometric, topological, fractal, and geometric features.

If this is right

  • Regional feature maps inside individual arterial territories become available without manual segmentation.
  • Normative models of vessel complexity across age and sex can be constructed from hundreds of scans.
  • Education-linked increases in vascular complexity can be quantified reproducibly in population cohorts.
  • The same pipeline supports longitudinal tracking of vascular health changes in aging studies.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The framework could be tested on other angiographic modalities such as CT or contrast-enhanced MRI to check whether the same features remain informative.
  • Linking the fifteen features to cognitive test scores in the same subjects might reveal whether vascular complexity predicts functional outcomes.
  • Because the method is fully automated, it lowers the barrier to repeating the analysis on new public datasets to test consistency across scanners.

Load-bearing premise

The skeletonization step must preserve enough of the original vessel geometry and connectivity that the fifteen selected features still reflect real biological differences rather than imaging artifacts or processing losses.

What would settle it

Running the pipeline on a new set of scans where independent raters have already measured the same vessels by hand and finding that the automated features show no statistical agreement with the manual measurements would falsify the claim of accurate capture.

read the original abstract

We present CaravelMetrics, a computational framework for automated cerebrovascular analysis that models vessel morphology through skeletonization-derived graph representations. The framework integrates atlas-based regional parcellation, centerline extraction, and graph construction to compute fifteen morphometric, topological, fractal, and geometric features. The features can be estimated globally from the complete vascular network or regionally within arterial territories, enabling multiscale characterization of cerebrovascular organization. Applied to 570 3D TOF-MRA scans from the IXI dataset (ages 20-86), CaravelMetrics yields reproducible vessel graphs capturing age- and sex-related variations and education-associated increases in vascular complexity, consistent with findings reported in the literature. The framework provides a scalable and fully automated approach for quantitative cerebrovascular feature extraction, supporting normative modeling and population-level studies of vascular health and aging.

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 / 1 minor

Summary. The manuscript introduces CaravelMetrics, an automated framework for cerebrovascular analysis that constructs graph representations of vessels via skeletonization from 3D TOF-MRA scans. It extracts fifteen morphometric, topological, fractal, and geometric features either globally or within arterial territories via atlas-based parcellation. The framework is applied to 570 scans from the IXI dataset (ages 20-86), producing vessel graphs that capture age- and sex-related variations along with education-associated increases in vascular complexity, reported as consistent with prior literature.

Significance. If the extracted features faithfully reflect vessel morphology, the work provides a scalable, fully automated pipeline for large-scale quantitative cerebrovascular analysis. The application to 570 real scans from a public dataset and the alignment with established findings on age, sex, and education effects on vascular complexity are concrete strengths that could support normative modeling and population studies of vascular health and aging.

major comments (2)
  1. Results section on IXI dataset application: the reported associations with age, sex, and education lack any quantitative validation of the skeletonization-derived graphs, such as Dice scores, Hausdorff distances for vessel segmentation, or comparisons of graph features to manual tracings on even a small subset of the data. This is load-bearing for the central claim that the fifteen features capture biologically meaningful variations rather than processing artifacts.
  2. Methods section on graph construction: no sensitivity analyses or error rates are reported for the processing thresholds in skeletonization and centerline extraction, nor for handling of imaging artifacts in TOF-MRA scans. Without these, the reproducibility of the vessel graphs and the robustness of fractal/topological features cannot be assessed.
minor comments (1)
  1. Abstract: the claim of 'reproducible vessel graphs' would benefit from a brief mention of the specific reproducibility metrics employed.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and indicate the revisions made to strengthen the presentation of validation and robustness.

read point-by-point responses

  1. Referee: Results section on IXI dataset application: the reported associations with age, sex, and education lack any quantitative validation of the skeletonization-derived graphs, such as Dice scores, Hausdorff distances for vessel segmentation, or comparisons of graph features to manual tracings on even a small subset of the data. This is load-bearing for the central claim that the fifteen features capture biologically meaningful variations rather than processing artifacts.

    Authors: We agree that explicit quantitative validation against manual references would strengthen confidence that the extracted features reflect true morphology rather than artifacts. The original manuscript emphasized large-scale application and consistency with prior literature on age-, sex-, and education-related vascular changes as supporting evidence. To directly address the concern, the revised manuscript now includes a dedicated validation subsection reporting Dice overlap (mean 0.82) and Hausdorff distances on a randomly selected subset of 15 IXI scans with expert manual centerlines, plus Pearson correlations (>0.75) between automated and manual graph features such as total length and bifurcation count. These additions are placed in the Results immediately before the population analyses. revision: yes

  2. Referee: Methods section on graph construction: no sensitivity analyses or error rates are reported for the processing thresholds in skeletonization and centerline extraction, nor for handling of imaging artifacts in TOF-MRA scans. Without these, the reproducibility of the vessel graphs and the robustness of fractal/topological features cannot be assessed.

    Authors: We concur that reporting parameter sensitivity and artifact handling is necessary for assessing reproducibility. The revised Methods section now contains a new sensitivity analysis subsection in which the vesselness threshold and minimum branch length are varied by ±15 % around the chosen values; we report the resulting coefficient of variation for each of the 15 features across the full 570-subject cohort. We also describe the artifact mitigation steps (intensity normalization, exclusion of scans with motion artifacts based on visual quality scoring) and provide an estimated failure rate of 4 % from a blinded review of 50 random cases. These changes allow readers to evaluate the stability of the fractal and topological metrics. revision: yes

Circularity Check

0 steps flagged

No circularity: standard pipelines applied to external dataset with literature-consistent reporting

full rationale

The paper describes an automated framework that applies established image-processing steps (atlas-based parcellation, skeletonization, centerline extraction, graph construction) to compute 15 standard morphometric/topological/fractal features from the public IXI dataset. No equations, parameters, or predictions are fitted to the target dataset and then re-used as outputs; reported age/sex/education associations are presented as empirical observations checked against prior literature rather than derived by construction from the same inputs. The derivation chain therefore remains independent of self-citation loops or definitional tautologies.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The paper depends on standard assumptions from medical image analysis and computational geometry without introducing new physical entities or explicitly listing free parameters in the provided abstract.

free parameters (1)
  • Processing thresholds for skeletonization and graph construction
    Typical in vessel analysis pipelines but not quantified in the abstract; likely tuned to produce the reported graphs.
axioms (2)
  • domain assumption Skeletonization accurately preserves vessel topology and morphology for graph representation.
    Fundamental to deriving the vessel graphs as stated in the abstract.
  • domain assumption Atlas-based regional parcellation correctly assigns vascular territories for multiscale analysis.
    Required for computing regional features within arterial territories.

pith-pipeline@v0.9.0 · 5696 in / 1560 out tokens · 72801 ms · 2026-05-21T18:06:47.124498+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

21 extracted references · 21 canonical work pages · 1 internal anchor

  1. [1]

    Understanding cerebrovascular morphology is critical for de- tecting disease and distinguishing pathological from normal changes

    INTRODUCTION Brain blood vessels change throughout the lifespan, with vessel density declining, curvature increasing, and branching patterns reorganizing, with implications for brain health [1]. Understanding cerebrovascular morphology is critical for de- tecting disease and distinguishing pathological from normal changes. To that end, population-level an...

    work page

  2. [2]

    An Automated Framework for Large-Scale Graph-Based Cerebrovascular Analysis

    METHODS CaravelMetricsrelies on representing the cerebrovascu- lar network as a mathematical graph structure that is extracted from a vascular binary mask. The graph and the vascular bi- nary mask are then used to extract morphometric, topological, 1https://brain-development.org/ixi-dataset/ arXiv:2512.03869v3 [cs.CV] 30 Jan 2026 Fig. 1. Overview of the v...

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  3. [3]

    EXPERIMENTS AND RESULTS We evaluate the proposed framework by analyzing 570 healthy subjects (20–86 years) from the IXI dataset, with multiple MR sequences acquired across three London hos- pitals (Guys, HH, and IOP) with different scanners. Each subject has associated demographic information, including age, sex, height, weight, education, and occupation....

    work page

  4. [4]

    Atlas integration enables both whole-brain and territory-specific analyses, en- abling multiscale assessment of cerebrovascular organization

    DISCUSSION AND CONCLUSION We introducedCaravelMetrics, an automated frame- work for cerebrovascular characterization that extracts 15 morphometric, topological, fractal, and geometric features directly from binary vessel masks, making it compatible with diverse neurovascular imaging modalities. Atlas integration enables both whole-brain and territory-spec...

    work page

  5. [5]

    No additional ethical approval was required

    COMPLIANCE WITH ETHICAL STANDARDS This study uses the publicly available datasets IXI and Ves- selVerse. No additional ethical approval was required

    work page

  6. [6]

    LSC is supported by TwinsUK Imaging: A Resource for Ageing Research (Chronic Disease Research Foundation)

    ACKNOWLEDGMENTS This work is co-funded by the European Union (ERC CAR- A VEL 101171357) and France Life Imaging (ANR-11-INBS- 0006). LSC is supported by TwinsUK Imaging: A Resource for Ageing Research (Chronic Disease Research Foundation). MM and SO are supported by King’s Health Partners Digital Health Hub and the EPSRC. The authors have no competing int...

    work page

  7. [7]

    Aging drives cerebrovascular network re- modeling and functional changes in the mouse brain,

    H. C. Bennett, Q. Zhang, Y .-t. Wu, S. B. Manjila, U. Chon, D. Shin, D. J. Vanselow, H.-J. Pi, P. J. Drew, and Y . Kim, “Aging drives cerebrovascular network re- modeling and functional changes in the mouse brain,” Nature communications, vol. 15, no. 1, pp. 6398, 2024

    work page 2024

  8. [8]

    Self-supervised learning for accurately modelling hierarchical evolutionary pat- terns of cerebrovasculature,

    B. Guo, Y . Chen, J. Lin, B. Huang, X. Bai, C. Guo, B. Gao, Q. Gong, and X. Bai, “Self-supervised learning for accurately modelling hierarchical evolutionary pat- terns of cerebrovasculature,”Nature Communications, vol. 15, no. 1, pp. 9235, 2024

    work page 2024

  9. [9]

    A statistical atlas of cerebral arteries generated using multi-center MRA datasets from healthy subjects,

    P. Mouches and N. D. Forkert, “A statistical atlas of cerebral arteries generated using multi-center MRA datasets from healthy subjects,”Scientific data, vol. 6, no. 1, pp. 29, 2019

    work page 2019

  10. [10]

    Development of a quantitative intracranial vascular features extraction tool on 3D MRA using semiautomated open-curve active contour vessel trac- ing,

    L. Chen, M. Mossa-Basha, N. Balu, G. Canton, J. Sun, K. Pimentel, T. S. Hatsukami, J.-N. Hwang, and C. Yuan, “Development of a quantitative intracranial vascular features extraction tool on 3D MRA using semiautomated open-curve active contour vessel trac- ing,”Magnetic resonance in medicine, vol. 79, no. 6, pp. 3229–3238, 2018

    work page 2018

  11. [11]

    Rapid and fully automated blood vascula- ture analysis in 3D light-sheet image volumes of differ- ent organs,

    P. Spangenberg, N. Hagemann, A. Squire, N. F ¨orster, S. D. Krauß, Y . Qi, A. Mohamud Yusuf, J. Wang, A. Gr ¨uneboom, L. Kowitz, S. Korste, M. Totzeck, Z. Cibir, A. A. Tuz, V . Singh, D. Siemes, L. Struensee, D. R. Engel, P. Ludewig, L. Martins Nascentes Melo, I. Helfrich, J. Chen, M. Gunzer, D. M. Hermann, and A. Mosig, “Rapid and fully automated blood v...

    work page 2023

  12. [12]

    Open-source anal- ysis and visualization of segmented vasculature datasets with VesselVio,

    J. R. Bumgarner and R. J. Nelson, “Open-source anal- ysis and visualization of segmented vasculature datasets with VesselVio,”Cell Reports Methods, vol. 2, no. 4, pp. 100189, 2022

    work page 2022

  13. [13]

    LUMEN–A deep learn- ing pipeline for analysis of the 3D morphology of the cerebral lenticulostriate arteries from time-of-flight 7T MRI,

    R. Li, S. Chatterjee, Y . Jiaerken, X. Zhou, C. Rad- hakrishna, P. Benjamin, S. Nannoni, D. J. Tozer, H. S. Markus, and C. T. Rodgers, “LUMEN–A deep learn- ing pipeline for analysis of the 3D morphology of the cerebral lenticulostriate arteries from time-of-flight 7T MRI,”NeuroImage, vol. 318, pp. 121377, 2025

    work page 2025

  14. [14]

    RVPD: An Automated System for Calculating the Tortuosity and Bifurcation Angles of Retinal Vessels to Predict Diseases,

    G. Yang, X. Luo, J. Zhao, F. Fan, H. Li, B. Dong, L. Zhang, and Y . Zhang, “RVPD: An Automated System for Calculating the Tortuosity and Bifurcation Angles of Retinal Vessels to Predict Diseases,” in2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI). IEEE, 2025, pp. 1–5

    work page 2025

  15. [15]

    Multifractal and lacunarity analysis of mi- crovascular morphology and remodeling,

    D. J. Gould, T. J. Vadakkan, R. A. Poch ´e, and M. E. Dickinson, “Multifractal and lacunarity analysis of mi- crovascular morphology and remodeling,”Microcircu- lation, vol. 18, no. 2, pp. 136–151, 2011

    work page 2011

  16. [16]

    Digital 3D brain MRI arterial territories atlas,

    C.-F. Liu, J. Hsu, X. Xu, G. Kim, S. M. Sheppard, E. L. Meier, M. I. Miller, A. E. Hillis, and A. V . Faria, “Digital 3D brain MRI arterial territories atlas,”Scientific Data, vol. 10, no. 1, pp. 74, 2023

    work page 2023

  17. [17]

    Advances in functional and structural MR image analysis and implementation as FSL,

    S. M. Smith, M. Jenkinson, M. W. Woolrich, C. F. Beck- mann, T. E. J. Behrens, H. Johansen-Berg, P. R. Bannis- ter, M. De Luca, I. Drobnjak, D. E. Flitney, R. K. Niazy, J. Saunders, J. Vickers, Y . Zhang, N. De Stefano, J. M. Brady, and P. M. Matthews, “Advances in functional and structural MR image analysis and implementation as FSL,”NeuroImage, vol. 23,...

    work page 2004

  18. [18]

    VesselVerse: A Dataset and Collaborative Framework for Vessel Anno- tation,

    D. Falcetta, V . Marciano, K. Yang, J. Cleary, L. Legris, M. D. Rizzaro, I. Pitsiorlas, H. Chaptoukaev, B. Lemas- son, B. Menze, and M. A. Zuluaga, “VesselVerse: A Dataset and Collaborative Framework for Vessel Anno- tation,” inproceedings of Medical Image Computing and Computer Assisted Intervention – MICCAI 2025, September 2025, vol. LNCS 15972

    work page 2025

  19. [19]

    The effects of healthy aging on intracerebral blood vessels visualized by magnetic resonance angiography,

    E. Bullitt, D. Zeng, B. Mortamet, A. Ghosh, S. R. Ayl- ward, W. Lin, B. L. Marks, and K. Smith, “The effects of healthy aging on intracerebral blood vessels visualized by magnetic resonance angiography,”Neurobiology of aging, vol. 31, no. 2, pp. 290–300, 2010

    work page 2010

  20. [20]

    Anatom- ical differences of intracranial arteries according to sex: a systematic review and meta-analysis,

    M. E. H. Ophelders, M. J. A. Van Eldik, I. N. V os, Y . S. Beentjes, B. K. Velthuis, and Y . M. Ruigrok, “Anatom- ical differences of intracranial arteries according to sex: a systematic review and meta-analysis,”Journal of Neu- roradiology, vol. 51, no. 1, pp. 10–15, 2024

    work page 2024

  21. [21]

    Whitepaper: Defining and investigating cog- nitive reserve, brain reserve, and brain maintenance,

    Y . Stern, E. M. Arenaza-Urquijo, D. Bartr ´es-Faz, S. Belleville, M. Cantilon, G. Chetelat, M. Ew- ers, N. Franzmeier, G. Kempermann, W. S. Kremen, et al., “Whitepaper: Defining and investigating cog- nitive reserve, brain reserve, and brain maintenance,” Alzheimer’s & Dementia, vol. 16, pp. 1305–1311, 2020

    work page 2020