Aneumo: A Large-Scale Comprehensive Synthetic Dataset of Aneurysm Hemodynamics

Chen Jiang; Chensen Lin; Feiyang Xiao; Jianchao Ge; Qimeng Wang; Taiwei Zhang; Xiansheng Wang; Xigui Li; Xin Guo; Yichi Zhang

REVIEW 5 major objections 5 minor 21 references

Aneumo: A Large-Scale Comprehensive Synthetic Dataset of Aneurysm Hemodynamics

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper introduces Aneumo, a dataset of 10,000 synthetic intracranial aneurysm models with CFD hemodynamics at eight steady-state flow rates, generated from 466 real aneurysm geometries.

desk verdict Potentially useful large synthetic aneurysm CFD dataset, but sloppy number reporting and unvalidated deformation realism need fixing before the field can trust it. read the letter →

arxiv 2501.09980 v1 pith:L4P5P7T5 submitted 2025-01-17 cs.CV cs.AIcs.LG

Xigui Li , Yuanye Zhou , Feiyang Xiao , Xin Guo , Yichi Zhang , Chen Jiang , Jianchao Ge , Xiansheng Wang

show 5 more authors

Qimeng Wang Taiwei Zhang Chensen Lin Yuan Cheng Yuan Qi

This is my paper · ORCID

classification cs.CVcs.AIcs.LG

keywords intracranialaneurysmhemodynamicscomputationalfluiddynamicssyntheticdatasetwallshearstressmorphologysegmentationmasksteady-stateflow

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper introduces Aneumo, a large synthetic dataset for studying blood flow in intracranial aneurysms. Starting from 466 real aneurysm models, the authors remove the aneurysm and then randomly deform the vessel wall to create 10,000 new geometries: 466 aneurysm-free controls and 9,534 deformed aneurysm models. For every geometry they run CFD simulations at eight steady-state flow rates, recording velocity, pressure, and wall shear stress, and they provide imaging-like segmentation masks aligned with each 3D model. If the synthetic shapes are physiologically plausible, the dataset gives researchers a resource for linking aneurysm morphology to hemodynamics and for training data-driven rupture-risk models.

What carries the argument

The load-bearing mechanism is the generative pipeline: take a real aneurysm model, repair and resect the aneurysm to obtain an aneurysm-free vessel, then apply a surface deformation whose amplitude is randomly chosen in the range [0.5, 1.0] at the former aneurysm site to synthesize varied aneurysm shapes. The deformed geometries are converted to imaging-like segmentation masks, meshed with polyhedral cells (minimum size 0.15 mm with ten boundary layers), and simulated with the finite volume method under steady-state mass-flow inlet conditions. This pipeline is what makes the scale of the dataset possible.

What would settle it

Measure clinically meaningful shape descriptors (dome-to-neck ratio, aspect ratio, maximum size, parent-vessel angle) on the 9,534 deformed models and compare them with a clinical cohort; if the synthetic shapes do not span or match the observed clinical distribution, the dataset's representativeness claim would be falsified. A second check is to use the eight flow-rate fields to predict a hemodynamic marker and compare the marker's relationship to rupture status observed in patient-specific CFD studies.

Watch

Extended reading notes

Core claim

The central claim is that a dataset of this scale—10,000 synthetic aneurysm geometries paired with hemodynamic fields—can be built from a modest set of real models and can support systematic study of aneurysm hemodynamics. The paper reports hemodynamic fields for every model at eight steady-state mass-flow rates from 0.001 to 0.004 kg/s, including velocity, pressure, and wall shear stress, plus NIfTI mask files that resemble medical images. It also reports numerical validation: grid-independence tests at a 0.15 mm minimum mesh size, CFL numbers below 1, and residual convergence for velocity and pressure, which are meant to establish that the simulated fields are reliable.

Load-bearing premise

The random deformation operation, with amplitude randomly chosen in [0.5, 1.0], is assumed to create aneurysm shapes that are physiologically plausible and representative of real aneurysms; the paper does not validate the resulting shape distribution against clinical data.

Editorial extensions

If this is right

Researchers can train and benchmark AI models that map aneurysm geometry directly to hemodynamic fields or rupture-related flow metrics.
Studies can compare across eight inflow conditions to see which hemodynamic markers are flow-rate-dependent and which are stable.
The aneurysm-free controls make it possible to quantify how the presence of a sac changes velocity, pressure, and wall shear stress on the same vessel.
The scale of the dataset supports data-driven surrogate models that approximate CFD results more cheaply than full simulation.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Editorial inference: because the deformation amplitudes are random rather than fit to clinical shape statistics, the morphological diversity of the 9,534 deformed models may not match the real distribution of aneurysm shapes; testing this would require shape-metric comparison against a clinical cohort.
Editorial inference: the mask files are generated from clean surface meshes, so they are unlikely to reproduce the noise, partial-volume blur, and slice thickness of clinical angiography; a segmentation model trained on these masks may need fine-tuning on real images.
Editorial inference: the steady-state Newtonian rigid-wall setup is a deliberate simplification; pulsatile flow and compliant walls could alter WSS distributions, so the dataset is best read as a controlled baseline for hemodynamic mechanism studies rather than a full patient-specific model.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The one thing to know: Aneumo is a genuinely useful resource in the making. As far as I can tell, no public dataset combines real aneurysm geometries, thousands of deformed synthetic variants, imaging-like masks, and multi-rate CFD fields at this scale. If the files on GitHub match the description, this will save people doing data-driven hemodynamics a lot of work. The individual steps are standard—resection, deformation, polyhedral meshing, OpenFOAM—but the combination is new and the scale is real. The 9,534 deformed models plus 466 aneurysm-free controls from 466 AneuX cases is exactly what you'd want for training morphology-to-flow surrogate models.

That said, the paper as written has several soft spots. Minor but embarrassing: the abstract says 466 real models while Data Records says 468; the flow rate list repeats 0.003 kg/s so "eight" steady-state rates is at most seven unique values; and the grid-independence numbers in the text contradict the Fig. 5 legend (text has 0.05 mm→150k and 0.25 mm→880k; legend has 0.05 mm→880k and 0.25 mm→150k). Also, "measured" is the wrong word for CFD output; those are simulated. All fixable, but they matter because the paper's authority depends on careful quality control.

The larger issue is that the physiological plausibility of the random deformations is asserted, not demonstrated. Deformation amplitudes in [0.5, 1.0] are chosen to "maintain physiological plausibility," and the only evidence is a volume change rate histogram. That is weak. You need at least a morphological comparison—dome height, neck diameter, aspect ratio, size ratio, vessel tortuosity—between the deformed models and the real AneuX cases or published clinical distributions. If the synthetic shapes drift outside the clinical manifold, the dataset's downstream value for clinical prediction shrinks. This is a substantive gap, not a cosmetic one.

For a reader in aneurysm CFD or ML for medical imaging, this paper is worth a serious look. It deserves peer review, but with major revision: fix the number inconsistencies, reword "measured," and add morphological validation of the deformation step. The dataset is the contribution; the prose should not get in the way.

Referee Report

5 major / 5 minor

Summary. The paper introduces Aneumo, a large synthetic dataset of intracranial aneurysm hemodynamics derived from real aneurysm models in the AneuX dataset. The authors describe a pipeline that resects aneurysms to create aneurysm-free vessel models, applies randomized deformations to generate 9,534 synthetic aneurysm models, converts the geometries to segmentation masks, generates polyhedral meshes, and runs CFD simulations at eight nominal steady-state mass flow rates between 0.001 and 0.004 kg/s. The dataset is hosted on GitHub and includes STL models, mesh files, NIfTI masks, and VTK hemodynamic fields (velocity, pressure, wall shear stress). The paper reports quality-control steps including mesh-independence analysis, CFL monitoring, and residual convergence.

Significance. If the dataset is delivered as described and the synthetic geometries are physiologically plausible, Aneumo would be a valuable open resource for aneurysm research: it uniquely combines real-derived geometries, large-scale synthetic deformation, imaging-like masks, and CFD hemodynamic fields at multiple flow rates, which is well suited for data-driven modeling and AI-based prediction studies. The public GitHub hosting and the explicit step-by-step description of the geometry-processing and simulation pipeline are positive features. However, the scientific value rests on two load-bearing assumptions: that the randomized deformation produces realistic aneurysm morphologies, and that the numerical and bookkeeping inconsistencies are resolved so that users can trust the dataset as delivered. These issues currently prevent me from endorsing the dataset without substantial revision.

major comments (5)

[3D Model Deformation] The deformation amplitudes are stated to be 'randomly selected within the range [0.5, 1.0] to ensure the diversity of each synthesized case while maintaining physiological plausibility.' Physiological plausibility is asserted but never demonstrated. The only quantitative support provided is the volume-change-rate histogram in Fig. 2, which shows that most volume changes are within expected ranges but says nothing about whether the deformed shapes resemble real aneurysms in clinically relevant morphological parameters such as dome height, neck diameter, aspect ratio, size ratio, or vessel tortuosity. This is load-bearing because the dataset's claimed usefulness for pathogenesis and clinical prediction depends on the realism of the 9,534 deformed models. I ask the authors to add a direct morphological comparison between the deformed synthetic models and the real AneuX models, or against published clinical distributions, and to discuss any cases where the deformation produces implausible shapes.
[Data Records] The number of real aneurysm models is inconsistent: the abstract and the Background & Summary say 466, while the Data Records section says '468 real aneurysm models.' The Data Records section also states '10,466 (real and synthetic) segmentation mask files,' but 468 real plus 10,000 synthetic equals 10,468, not 10,466. Since the 466/468 count is the first concrete number a reader checks, this inconsistency undermines confidence in the dataset's quality control. The authors must reconcile these counts and state the exact number of real models, synthetic models, and mask files consistently across the abstract, main text, and repository documentation.
[Boundary Condition Definition and Hemodynamic Simulation] The list of mass flow rates contains a duplicate: '0.0030 kg/s, 0.003 kg/s' are the same value. The paper claims eight steady-state flow rates, but the list as printed has at most seven unique values. Because the abstract and the data records rely on the 'eight steady-state flow rates' claim, this error affects a central quantitative assertion. The authors should correct the flow-rate list, or if there truly are eight distinct rates, list them unambiguously.
[Technical Validation] The mesh-independence results in the text and in Fig. 5 contradict each other. The text states that minimum grid sizes of 0.05 mm, 0.10 mm, 0.15 mm, 0.20 mm, and 0.25 mm correspond to total grid numbers of approximately 150,000, 250,000, 330,000, 330,000, and 880,000, respectively. The Fig. 5 legend, however, lists 0.05 mm (880,000), 0.10 mm (530,000), 0.15 mm (330,000), 0.20 mm (250,000), and 0.25 mm (150,000). These two sets of numbers cannot both be correct. Since the conclusion that 'a minimum grid size of 0.15 mm' gives a relative error of less than 1% depends on this comparison, the inconsistency is load-bearing for the reliability of the CFD data. The authors must correct the text or the figure and provide a consistent mesh-convergence table.
[Boundary Condition Definition and Hemodynamic Simulation] The manuscript states that the icoFoam solver and the PISO algorithm were used for steady-state simulations. icoFoam is a transient incompressible Navier-Stokes solver, and PISO is a time-accurate pressure-velocity coupling method; for genuinely steady-state simulations one would normally use a steady solver such as SIMPLE. The paper also reports maintaining CFL below 1, which is a time-step constraint for transient solvers. This raises a technical question about whether the reported fields are truly steady-state solutions or time-averaged/instantaneous snapshots of a transient run. The authors should clarify the solver configuration, the convergence criterion used to declare a steady state, and how the reported velocity, pressure, and WSS values were extracted.

minor comments (5)

[Abstract] The phrase 'hemodynamic data measured at eight steady-state flow rates' is inaccurate: the data are obtained from CFD simulations, not from measurements. The wording should be revised to 'simulated' or 'computed' throughout the manuscript where appropriate.
[3D Model Deformation] The word 'Ramesh' appears twice ('application of Ramesh operations' and 'Ramesh and Fit Surface optimizations'); this should presumably be 'remeshing.'
[Fig. 5] The x-axis label 'Dianstance/mm' contains a typo; it should be 'Distance/mm.'
[Author Contributions] The author name 'Taiwei, Zhang' has an unnecessary comma; it should be 'Taiwei Zhang' for consistency with the other author names.
[Code availability] The statement 'No custom code was utilized in this study' is unusual for a dataset paper. The deformation, meshing, and simulation steps rely on commercial software with many parameters; providing at least the boundary-condition settings, solver dictionaries, and postprocessing scripts would substantially improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the dataset is generated by an external-geometry pipeline plus CFD simulation, with no fitted parameter renamed as a prediction.

full rationale

The paper's construction chain is self-contained and non-circular: it starts from 466 real aneurysm models taken from the external AneuX dataset, applies aneurysm resection and randomized deformation, converts the resulting geometries to masks, generates meshes, and runs OpenFOAM CFD simulations under stated boundary conditions. The claimed outputs (10,000 synthetic models, segmentation masks, and hemodynamic fields at multiple flow rates) are produced by this pipeline rather than being defined in terms of the dataset's own claims. No equation in the paper defines an output quantity as equivalent to an input parameter, and no parameter is fitted to a subset of the delivered data and then reported as a prediction of a closely related quantity. The only citation of prior work is the use of AneuX as the source of real geometries, which is an external dataset and not authored by the present authors; no uniqueness theorem or ansatz is imported from the authors' own prior work. The paper's assertion that deformation amplitudes in [0.5, 1.0] maintain 'physiological plausibility' is an unvalidated assumption, which is a correctness or validation concern rather than a circularity, because the plausibility claim is not used to derive the deformation operation or the CFD outputs. The internal inconsistencies noted in the review (duplicated mass-flow value, contradictory grid-count numbers, and 468 vs. 466 real models) affect reproducibility and quality reporting but do not make any derived quantity equivalent to its input by construction. Accordingly, no circular step is present.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The dataset depends entirely on modeling assumptions about blood rheology, wall mechanics, steady-state flow, and the physiological plausibility of the deformation process. None of these are validated against clinical or experimental data, so the dataset's real-world relevance rests on domain assumptions that are plausible but unverified.

assumptions (5)

domain assumption Blood is modeled as an incompressible Newtonian fluid with density 1050 kg/m3 and dynamic viscosity 0.00345 Pa·s.
Invoked in Boundary Condition Definition; ignores non-Newtonian behavior of blood, potentially affecting wall shear stress in low-shear regions.
domain assumption Vessel walls are rigid no-slip walls.
Invoked in Boundary Condition Definition; ignores vessel wall compliance and fluid-structure interaction, which may be relevant in aneurysm mechanics.
domain assumption Steady-state flow at eight constant mass flow rates adequately captures hemodynamics relevant to aneurysms.
Used throughout; pulsatile flow effects are absent, limiting physiological fidelity and comparability to cardiac-cycle-dependent studies.
ad hoc to paper Random deformation amplitudes in [0.5, 1.0] yield physiologically plausible aneurysm morphologies.
Stated in 3D Model Deformation; no comparison to clinical shape distributions is provided, making this an unsupported assumption central to dataset usefulness.
domain assumption The cross-section with the largest pipe diameter is chosen as the inlet because AneuX lacks imaging data.
Stated in Mesh Generation; this heuristic may not correspond to the true physiological inlet direction, potentially affecting downstream flow solutions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Aneumo: A Large-Scale Comprehensive Synthetic Dataset of Aneurysm Hemodynamics." pith.science (2026). https://pith.science/paper/L4P5P7T5

@misc{pith2026250109980,
  author       = {Pith},
  title        = {Pith review of: Aneumo: A Large-Scale Comprehensive Synthetic Dataset of Aneurysm Hemodynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L4P5P7T5}},
  note         = {Machine review of arXiv:2501.09980}
}

read the original abstract

Intracranial aneurysm (IA) is a common cerebrovascular disease that is usually asymptomatic but may cause severe subarachnoid hemorrhage (SAH) if ruptured. Although clinical practice is usually based on individual factors and morphological features of the aneurysm, its pathophysiology and hemodynamic mechanisms remain controversial. To address the limitations of current research, this study constructed a comprehensive hemodynamic dataset of intracranial aneurysms. The dataset is based on 466 real aneurysm models, and 10,000 synthetic models were generated by resection and deformation operations, including 466 aneurysm-free models and 9,534 deformed aneurysm models. The dataset also provides medical image-like segmentation mask files to support insightful analysis. In addition, the dataset contains hemodynamic data measured at eight steady-state flow rates (0.001 to 0.004 kg/s), including critical parameters such as flow velocity, pressure, and wall shear stress, providing a valuable resource for investigating aneurysm pathogenesis and clinical prediction. This dataset will help advance the understanding of the pathologic features and hemodynamic mechanisms of intracranial aneurysms and support in-depth research in related fields. Dataset hosted at https://github.com/Xigui-Li/Aneumo.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 20 canonical work pages

[1]

& Jafari, A

Zakeri, M., Atef, A., Aziznia, M. & Jafari, A. A comprehensive investigation of morphological features responsible for cerebr al aneurysm rupture using machine learning. Sci. Rep. 14, 15777 (2024)

work page 2024
[2]

& Rinkel, G

Etminan, N. & Rinkel, G. J. Unruptured intracranial aneurysms: development, rupture and preventive management. Nat. Rev. Neurol. 12, 699-713 (2016)

work page 2016
[3]

Chalouhi, N., Hoh, B. L. & Hasan, D. Review of Cerebral Aneurysm Formation, Growth, and Rupture. Stroke 44, 3613-3622 (2013)

work page 2013
[4]

Multidisciplinary Consensus on Assessment of Unruptured Intracranial Aneurysms

Etminan, N., et al. Multidisciplinary Consensus on Assessment of Unruptured Intracranial Aneurysms. Stroke 45, 1523-1530 (2014)

work page 2014
[5]

Steiner, T. , et al. European Stroke Organization Guidelines for the Management of Intracranial Aneurysms and Subarachnoid Haemorrhage. Cerebrovascular Diseases 35, 93-112 (2013)

work page 2013
[6]

S., Turjman, F

Turjman, A. S., Turjman, F. & Edelman, E. R. Role of Fluid Dynamics and Inflammation in Intracranial Aneurysm Formation. Circulation 129, 373-382 (2014)

work page 2014
[7]

M., Putman, C

Sforza, D. M., Putman, C. M. & Cebral, J. R. Hemodynamics of Cerebral Aneurysms. Annu. Rev. Fluid Mech. 41, 91-107 (2009)

work page 2009
[8]

C., Luo, C., Li, Y

Chen, J. C., Luo, C., Li, Y. & Tan, D. H. Knowledge domain and emerging trends in the rupture risk of intracranial aneurysms research from 2004 to 2023. World J Clin Cases 12, 5382-5403 (2024)

work page 2024

Show all 21 references

[9]

Detection and analysis of cerebral aneurysms based on X-ray rotational angiography - the CADA 2020 challenge

Ivantsits, M., et al. Detection and analysis of cerebral aneurysms based on X-ray rotational angiography - the CADA 2020 challenge. Med. Image Anal. 77, 102333 (2022)

2022
[10]

Di Noto, T. , et al. Towards Automated Brain Aneurysm Detection in TOF -MRA: Open Data, Weak Labels, and Anatomical Knowledge. Neuroinformatics 21, 21-34 (2023)

2023
[11]

Timmins, K. M. , et al. Comparing methods of detecting and segmenting unruptured intracranial aneurysms on TOF -MRAS: The ADAM challenge. NeuroImage 238, 118216 (2021)

2021
[12]

& Hirsch, S

Juchler, N., Schilling, S., Bijlenga, P., Kurtcuoglu, V. & Hirsch, S. Shape Trumps Size: Image -Based Morphological Analysis Reveals That the 3D Shape Discriminates Intracranial Aneurysm Disease Status Better Than Aneurysm Size. Front. Neurol. 13 (2022)

2022
[13]

& Igarashi, T

Yang, X., Xia, D., Kin, T. & Igarashi, T. in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2653-2663

2020
[14]

Intracranial aneurysm CTA images and 3D models dataset with clinical morphological and hemodynamic data

Song, M., et al. Intracranial aneurysm CTA images and 3D models dataset with clinical morphological and hemodynamic data. Sci. Data 11, 1213 (2024)

2024
[15]

& Dashti, R

Gilotra, K., Swarna, S., Mani, R., Basem, J. & Dashti, R. Role of artificial intelligence and machine learning in the diagnos is of cerebrovascular disease. Front. Hum. Neurosci. 17 (2023)

2023
[16]

Recent advances on machine learning for computational fluid dynamics: A survey

Wang, H., et al. Recent advances on machine learning for computational fluid dynamics: A survey. arXiv preprint arXiv:2408.12171 (2024)

2024 arXiv
[17]

& Brunton, S

Vinuesa, R. & Brunton, S. L. Enhancing computational fluid dynamics with machine learning. Nat. Comput. Sci. 2, 358-366 (2022)

2022
[18]

Machine learning–accelerated computational fluid dynamics

Kochkov, D., et al. Machine learning–accelerated computational fluid dynamics. PNAS 118, e2101784118 (2021)

2021
[19]

Vinuesa, R., Brunton, S. L. & McKeon, B. J. The transformative potential of machine learning for experiments in fluid mechani cs. Nat. Rev. Phys. 5, 536-545 (2023)

2023
[20]

& Okaze, T

Wang, W., Cao, Y. & Okaze, T. Comparison of hexahedral, tetrahedral and polyhedral cells for reproducing the wind field around an isolated building by LES. Build. Environ. 195, 107717 (2021)

2021
[21]

K., et al

Paritala, P. K., et al. Reproducibility of the computational fluid dynamic analysis of a cerebral aneurysm monitored over a decade. Sci. Rep. 13, 219 (2023). Acknowledgments This paper was supported by the National Natural Science Foundation of China ( Grant No. 82394432 and 9...

2023

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

& Jafari, A

Zakeri, M., Atef, A., Aziznia, M. & Jafari, A. A comprehensive investigation of morphological features responsible for cerebr al aneurysm rupture using machine learning. Sci. Rep. 14, 15777 (2024)

work page 2024

[2] [2]

& Rinkel, G

Etminan, N. & Rinkel, G. J. Unruptured intracranial aneurysms: development, rupture and preventive management. Nat. Rev. Neurol. 12, 699-713 (2016)

work page 2016

[3] [3]

Chalouhi, N., Hoh, B. L. & Hasan, D. Review of Cerebral Aneurysm Formation, Growth, and Rupture. Stroke 44, 3613-3622 (2013)

work page 2013

[4] [4]

Multidisciplinary Consensus on Assessment of Unruptured Intracranial Aneurysms

Etminan, N., et al. Multidisciplinary Consensus on Assessment of Unruptured Intracranial Aneurysms. Stroke 45, 1523-1530 (2014)

work page 2014

[5] [5]

Steiner, T. , et al. European Stroke Organization Guidelines for the Management of Intracranial Aneurysms and Subarachnoid Haemorrhage. Cerebrovascular Diseases 35, 93-112 (2013)

work page 2013

[6] [6]

S., Turjman, F

Turjman, A. S., Turjman, F. & Edelman, E. R. Role of Fluid Dynamics and Inflammation in Intracranial Aneurysm Formation. Circulation 129, 373-382 (2014)

work page 2014

[7] [7]

M., Putman, C

Sforza, D. M., Putman, C. M. & Cebral, J. R. Hemodynamics of Cerebral Aneurysms. Annu. Rev. Fluid Mech. 41, 91-107 (2009)

work page 2009

[8] [8]

C., Luo, C., Li, Y

Chen, J. C., Luo, C., Li, Y. & Tan, D. H. Knowledge domain and emerging trends in the rupture risk of intracranial aneurysms research from 2004 to 2023. World J Clin Cases 12, 5382-5403 (2024)

work page 2024