pith. sign in

arxiv: 2411.09593 · v3 · pith:DWD3JNPCnew · submitted 2024-11-14 · 📡 eess.IV · cs.AI· cs.CV

SMILE-UHURA Challenge -- Small Vessel Segmentation at Mesoscopic Scale from Ultra-High Resolution 7T Magnetic Resonance Angiograms

Soumick Chatterjee , Hendrik Mattern , Marc D\"orner , Alessandro Sciarra , Florian Dubost , Hannes Schnurre , Rupali Khatun , Chun-Chih Yu
show 40 more authors
Tsung-Lin Hsieh Yi-Shan Tsai Yi-Zeng Fang Yung-Ching Yang Juinn-Dar Huang Marshall Xu Siyu Liu Fernanda L. Ribeiro Saskia Bollmann Karthikesh Varma Chintalapati Chethan Mysuru Radhakrishna Sri Chandana Hudukula Ram Kumara Raviteja Sutrave Abdul Qayyum Moona Mazher Imran Razzak Cristobal Rodero Steven Niederren Fengming Lin Yan Xia Jiacheng Wang Riyu Qiu Liansheng Wang Arya Yazdan Panah Rosana El Jurdi Guanghui Fu Janan Arslan Ghislain Vaillant Romain Valabregue Didier Dormont Bruno Stankoff Olivier Colliot Luisa Vargas Isai Daniel Chac\'on Ioannis Pitsiorlas Pablo Arbel\'aez Maria A. Zuluaga Stefanie Schreiber Oliver Speck Andreas N\"urnberger
This is my paper

Pith reviewed 2026-05-23 17:39 UTC · model grok-4.3

classification 📡 eess.IV cs.AIcs.CV
keywords small vessel segmentation7T MRImagnetic resonance angiographydeep learningchallengeToF MRAcerebral small vessel diseaseimage segmentation
0
0 comments X

The pith

A new annotated 7T MRA dataset lets deep learning methods segment small brain vessels with Dice scores up to 0.838.

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

Small vessels at the mesoscopic scale in the brain are difficult to image and segment, which limits research into cerebral small vessel diseases, but higher-resolution 7T MRI now makes them visible. The SMILE-UHURA challenge supplies the first public annotated dataset of 7T Time-of-Flight angiograms, built by combining automated pre-segmentation with extensive manual refinement. Sixteen submitted deep learning methods plus two baselines were tested on both held-out volumes from the training distribution and a fully secret external 7T dataset. Most methods reached reliable performance, with peak Dice scores of 0.838 on internal tests and 0.716 on the external set. This outcome shows that the released data and challenge framework support development of practical segmentation tools for these vessels.

Core claim

The SMILE-UHURA challenge shows that deep learning methods trained on its annotated 7T ToF MRA dataset achieve reliable segmentation of small cerebral vessels, attaining Dice scores up to 0.838 ± 0.066 on held-out test data from the same source and 0.716 ± 0.125 on a separate secret 7T dataset, with average performance reaching 0.804 ± 0.15.

What carries the argument

The SMILE-UHURA annotated dataset of ultra-high-resolution 7T Time-of-Flight MRAs created via automated pre-segmentation followed by manual refinement.

If this is right

  • Deep learning models can be trained and benchmarked for mesoscopic vessel segmentation using publicly released labels.
  • Current architectures handle the spatial complexity of small vessels visible at 7T resolution with usable accuracy.
  • Performance holds on both internal held-out data and an independent external dataset, indicating some robustness.
  • The dataset directly supports quantitative studies of pathologies that affect small cerebral vessels.

Where Pith is reading between the lines

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

  • The lower external-dataset scores suggest that domain adaptation or additional training data could further improve generalization.
  • The dataset could be combined with other modalities to study vessel changes linked to cognitive decline.
  • Releasing the pre-segmentation code alongside the labels would let others replicate and extend the annotation pipeline.
  • Multi-rater label sets on a subset of volumes would allow uncertainty-aware training and better error estimation.

Load-bearing premise

The manual refinement step after automated pre-segmentation produced accurate ground-truth labels that generalize to unseen 7T volumes.

What would settle it

Independent expert segmentations of the secret test volumes that produce Dice scores for the top methods substantially below the reported values would falsify the claim of reliable performance.

Figures

Figures reproduced from arXiv: 2411.09593 by Abdul Qayyum, Alessandro Sciarra, Andreas N\"urnberger, Arya Yazdan Panah, Bruno Stankoff, Chethan Mysuru Radhakrishna, Chun-Chih Yu, Cristobal Rodero, Didier Dormont, Fengming Lin, Fernanda L. Ribeiro, Florian Dubost, Ghislain Vaillant, Guanghui Fu, Hannes Schnurre, Hendrik Mattern, Imran Razzak, Ioannis Pitsiorlas, Isai Daniel Chac\'on, Janan Arslan, Jiacheng Wang, Juinn-Dar Huang, Karthikesh Varma Chintalapati, Liansheng Wang, Luisa Vargas, Marc D\"orner, Maria A. Zuluaga, Marshall Xu, Moona Mazher, Oliver Speck, Olivier Colliot, Pablo Arbel\'aez, Raviteja Sutrave, Riyu Qiu, Romain Valabregue, Rosana El Jurdi, Rupali Khatun, Saskia Bollmann, Siyu Liu, Soumick Chatterjee, Sri Chandana Hudukula Ram Kumara, Stefanie Schreiber, Steven Niederren, Tsung-Lin Hsieh, Yan Xia, Yi-Shan Tsai, Yi-Zeng Fang, Yung-Ching Yang.

Figure 1
Figure 1. Figure 1: Dice scores on the test subset from the open dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Jaccard index scores on the test subset from the open dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Volumetric similarity coefficients on the test subset from the open dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). 14 [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Mutual information scores on the test subset from the open dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Balanced average Hausdorff distances on the test subset from the open dataset. The plot was confined to bAHD <= 3, as three methods yielded extreme values, thereby rendering the remainder of the comparisons incomprehensible. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). 16 [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Dice scores on the secret dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., UNet MSS). [PITH_FULL_IMAGE:figures/full_fig_p019_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Jaccard index scores on the secret dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., UNet MSS). [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Volumetric similarity coefficients on the secret dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). 21 [PITH_FULL_IMAGE:figures/full_fig_p021_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Mutual information scores on the secret dataset. The red dashed line denotes the median of the better-performing baseline method (i.e., UNet MSS). [PITH_FULL_IMAGE:figures/full_fig_p022_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Balanced average Hausdorff distances on the secret dataset. The plot was confined to bAHD <= 3, as six methods yielded extreme values, thereby rendering the remainder of the comparisons incomprehensible. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). 23 [PITH_FULL_IMAGE:figures/full_fig_p023_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Dice scores on both the datasets combined. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). [PITH_FULL_IMAGE:figures/full_fig_p025_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Jaccard index scores on both the datasets combined. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). [PITH_FULL_IMAGE:figures/full_fig_p026_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Volumetric similarity coefficients on both the datasets combined. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). 27 [PITH_FULL_IMAGE:figures/full_fig_p027_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Mutual information scores on both the datasets combined. The red dashed line denotes the median of the better-performing baseline method (i.e., UNet [PITH_FULL_IMAGE:figures/full_fig_p028_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Balanced average Hausdorff distances on both the datasets combined. The plot was confined to bAHD <= 3, as eight methods yielded extreme values, thereby rendering the remainder of the comparisons incomprehensible. The red dashed line denotes the median of the better-performing baseline method (i.e., DS6). 29 [PITH_FULL_IMAGE:figures/full_fig_p029_15.png] view at source ↗
read the original abstract

The human brain receives nutrients and oxygen through an intricate network of blood vessels. Pathology affecting small vessels, at the mesoscopic scale, represents a critical vulnerability within the cerebral blood supply and can lead to severe conditions, such as Cerebral Small Vessel Diseases. The advent of 7 Tesla MRI systems has enabled the acquisition of higher spatial resolution images, making it possible to visualise such vessels in the brain. However, the lack of publicly available annotated datasets has impeded the development of robust, machine learning-driven segmentation algorithms. To address this, the SMILE-UHURA challenge was organised. This challenge, held in conjunction with the ISBI 2023, in Cartagena de Indias, Colombia, aimed to provide a platform for researchers working on related topics. The SMILE-UHURA challenge addresses the gap in publicly available annotated datasets by providing an annotated dataset of Time-of-Flight angiography acquired with 7T MRI. This dataset was created through a combination of automated pre-segmentation and extensive manual refinement. In this manuscript, sixteen submitted methods and two baseline methods are compared both quantitatively and qualitatively on two different datasets: held-out test MRAs from the same dataset as the training data (with labels kept secret) and a separate 7T ToF MRA dataset where both input volumes and labels are kept secret. The results demonstrate that most of the submitted deep learning methods, trained on the provided training dataset, achieved reliable segmentation performance. Dice scores reached up to 0.838 $\pm$ 0.066 and 0.716 $\pm$ 0.125 on the respective datasets, with an average performance of up to 0.804 $\pm$ 0.15.

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

1 major / 2 minor

Summary. The paper presents the SMILE-UHURA challenge, which supplies a publicly available annotated dataset of 7T ToF MRAs for mesoscopic-scale small vessel segmentation. The dataset was generated via automated pre-segmentation followed by manual refinement. Sixteen submitted deep-learning methods plus two baselines are evaluated quantitatively and qualitatively on a held-out test set from the same distribution (labels secret) and on a fully secret external 7T dataset; the central claim is that most methods achieved reliable performance, with peak Dice scores of 0.838 ± 0.066 and 0.716 ± 0.125 respectively.

Significance. If the ground-truth labels prove accurate, the work supplies the first public benchmark for 7T small-vessel segmentation and demonstrates that current DL architectures can generalize to an external secret test volume. The dual-test design (internal held-out plus external secret) is a strength that directly supports claims of practical utility for cerebral small-vessel disease studies.

major comments (1)
  1. [Methods / dataset creation] Dataset creation (Methods section describing annotation): the protocol of automated pre-segmentation plus manual refinement is presented without any quantitative inter-rater agreement metrics (Dice, Hausdorff, or boundary variability), without comparison to a second independent expert, and without sensitivity analysis on ambiguous small-vessel boundaries. Because the central claim of “reliable segmentation performance” rests on these labels being faithful to anatomy, the absence of such validation directly weakens the interpretation of all reported Dice scores.
minor comments (2)
  1. [Abstract and Results] The abstract and results tables report mean ± std but do not state the number of test volumes or whether the std is across volumes or across methods; adding this information would improve clarity of the performance claims.
  2. [Introduction / Discussion] No reference is made to existing small-vessel segmentation benchmarks (e.g., other 7T or high-resolution angiography datasets) that could contextualize the achieved Dice values.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on the annotation protocol. We address the single major comment point-by-point below.

read point-by-point responses

  1. Referee: [Methods / dataset creation] Dataset creation (Methods section describing annotation): the protocol of automated pre-segmentation plus manual refinement is presented without any quantitative inter-rater agreement metrics (Dice, Hausdorff, or boundary variability), without comparison to a second independent expert, and without sensitivity analysis on ambiguous small-vessel boundaries. Because the central claim of “reliable segmentation performance” rests on these labels being faithful to anatomy, the absence of such validation directly weakens the interpretation of all reported Dice scores.

    Authors: We agree that the absence of quantitative inter-rater metrics is a limitation. The labels were produced by automated pre-segmentation followed by manual refinement performed by a single experienced rater; no second independent expert annotation or inter-rater statistics were collected during dataset creation. We will revise the Methods section to expand the description of the annotation workflow and add an explicit limitations paragraph in the Discussion that states this limitation and its potential effect on the interpretation of the reported Dice scores (0.838 internal, 0.716 external). This revision will be made without changing any numerical results or claims about method performance. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical challenge benchmark with no derivations or fitted predictions

full rationale

The manuscript is a report on a segmentation challenge. It describes dataset creation via automated pre-segmentation plus manual refinement, then reports Dice scores of submitted DL methods on held-out test volumes (labels secret) and an external secret dataset. No equations, first-principles derivations, parameter fits, or predictions are presented that could reduce to the inputs by construction. No self-citation chains or uniqueness theorems are invoked to support any claim. The evaluation is self-contained against external benchmarks (the challenge test sets), satisfying the criteria for a non-circular empirical report.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an empirical challenge report; no mathematical derivations, fitted parameters, or new postulated entities are present.

pith-pipeline@v0.9.0 · 6097 in / 1233 out tokens · 27059 ms · 2026-05-23T17:39:37.909123+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

46 extracted references · 46 canonical work pages · 1 internal anchor

  1. [1]

    J. M. Wardlaw, C. Smith, M. Dichgans, Small vessel disease: mechanisms and clinical implications, The Lancet Neurology 18 (7) (2019) 684–696

    work page 2019

  2. [2]

    Chalkias, I.-N

    E. Chalkias, I.-N. Chalkias, C. Bakirtzis, L. Messinis, G. Nasios, P. Ioan- nidis, D. Pirounides, Differentiating degenerative from vascular dementia with the help of optical coherence tomography angiography biomarkers, in: Healthcare, V ol. 10, MDPI, 2022, p. 539

    work page 2022

  3. [3]

    deep learning system

    Y . Duan, W. Shan, L. Liu, Q. Wang, Z. Wu, P. Liu, J. Ji, Y . Liu, K. He, Y . Wang, Primary categorizing and masking cerebral small vessel disease based on “deep learning system”, Frontiers in Neuroinformatics 14 (2020) 17. 33

    work page 2020

  4. [4]

    Litak, M

    J. Litak, M. Mazurek, B. Kulesza, P. Szmygin, J. Litak, P. Kamieniak, C. Grochowski, Cerebral small vessel disease, International journal of molecular sciences 21 (24) (2020) 9729

    work page 2020

  5. [5]

    Hendrikse, J

    J. Hendrikse, J. J. Zwanenburg, F. Visser, T. Takahara, P. Luijten, Non- invasive depiction of the lenticulostriate arteries with time-of-flight mr angiography at 7.0 t, Cerebrovascular diseases 26 (6) (2008) 624–629

    work page 2008

  6. [6]

    Kang, C.-W

    C.-K. Kang, C.-W. Park, J.-Y . Han, S.-H. Kim, C.-A. Park, K.-N. Kim, S.-M. Hong, Y .-B. Kim, K. H. Lee, Z.-H. Cho, Imaging and analysis of lenticulostriate arteries using 7.0-tesla magnetic resonance angiography, Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine 61 (1) (2009) 136–144

    work page 2009

  7. [7]

    Mattern, A

    H. Mattern, A. Sciarra, F. Godenschweger, D. Stucht, F. L ¨usebrink, G. Rose, O. Speck, Prospective motion correction enables highest res- olution time-of-flight angiography at 7t, Magnetic resonance in medicine 80 (1) (2018) 248–258

    work page 2018

  8. [8]

    Bollmann, H

    S. Bollmann, H. Mattern, M. Bernier, S. D. Robinson, D. Park, O. Speck, J. R. Polimeni, Imaging of the pial arterial vasculature of the human brain in vivo using high-resolution 7t time-of-flight angiography, Elife 11 (2022) e71186

    work page 2022

  9. [9]

    Staal, M

    J. Staal, M. D. Abr `amoff, M. Niemeijer, M. A. Viergever, B. Van Gin- neken, Ridge-based vessel segmentation in color images of the retina, IEEE transactions on medical imaging 23 (4) (2004) 501–509

    work page 2004

  10. [10]

    Al-Rawi, M

    M. Al-Rawi, M. Qutaishat, M. Arrar, An improved matched filter for blood vessel detection of digital retinal images, Computers in biology and medicine 37 (2) (2007) 262–267

    work page 2007

  11. [11]

    Annunziata, A

    R. Annunziata, A. Garzelli, L. Ballerini, A. Mecocci, E. Trucco, Leverag- ing multiscale hessian-based enhancement with a novel exudate inpaint- ing technique for retinal vessel segmentation, IEEE journal of biomedical and health informatics 20 (4) (2015) 1129–1138

    work page 2015

  12. [12]

    R. D. Rudyanto, S. Kerkstra, E. M. Van Rikxoort, C. Fetita, P.-Y . Brillet, C. Lefevre, W. Xue, X. Zhu, J. Liang, I. ¨Oks¨uz, et al., Comparing algo- rithms for automated vessel segmentation in computed tomography scans of the lung: the vessel12 study, Medical image analysis 18 (7) (2014) 1217–1232

    work page 2014

  13. [13]

    K. M. Timmins, I. C. van der Schaaf, E. Bennink, Y . M. Ruigrok, X. An, M. Baumgartner, P. Bourdon, R. De Feo, T. Di Noto, F. Dubost, et al., Comparing methods of detecting and segmenting unruptured intracranial aneurysms on tof-mras: The adam challenge, Neuroimage 238 (2021) 118216

    work page 2021

  14. [14]

    C. H. Sudre, K. Van Wijnen, F. Dubost, H. Adams, D. Atkinson, F. Barkhof, M. A. Birhanu, E. E. Bron, R. Camarasa, N. Chaturvedi, et al., Where is valdo? vascular lesions detection and segmentation challenge at miccai 2021, arXiv preprint arXiv:2208.07167

    work page arXiv 2021

  15. [15]

    W. Liao, K. Rohr, C. K. Kang, Z. H. Cho, L. Fellow, S. W ¨orz, Automatic 3D segmentation and quantification of lenticulostriate arteries from high- resolution 7 tesla MRA images, IEEE Transactions on Image Processing 25 (1) (2016) 400–413. doi:10.1109/TIP.2015.2499085

    work page doi:10.1109/tip.2015.2499085 2016

  16. [16]

    A. F. Frangi, W. J. Niessen, K. L. Vincken, M. A. Viergever, Multiscale vessel enhancement filtering, in: International conference on medical im- age computing and computer-assisted intervention, Springer, 1998, pp. 130–137

    work page 1998

  17. [17]

    Canero, P

    C. Canero, P. Radeva, Vesselness enhancement di ffusion, Pattern Recog- nition Letters 24 (16) (2003) 3141–3151

    work page 2003

  18. [18]

    Manniesing, W

    R. Manniesing, W. Niessen, Multiscale vessel enhancing di ffusion in ct angiography noise filtering, in: Biennial International Conference on In- formation Processing in Medical Imaging, Springer, 2005, pp. 138–149

    work page 2005

  19. [19]

    In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F

    O. Ronneberger, P. Fischer, T. Brox, U-net: Convolutional networks for biomedical image segmentation, in: International Conference on Medical image computing and computer-assisted intervention, Springer, 2015, pp. 234–241. doi:10.1007/978-3-319-24574-4\_28

    work page doi:10.1007/978-3-319-24574-4 2015

  20. [20]

    S. Yang, J. Kweon, J.-H. Roh, J.-H. Lee, H. Kang, L.-J. Park, D. J. Kim, H. Yang, J. Hur, D.-Y . Kang, et al., Deep learning segmentation of major vessels in x-ray coronary angiography, Scientific reports 9 (1) (2019) 1– 11

    work page 2019

  21. [21]

    Livne, J

    M. Livne, J. Rieger, O. U. Aydin, A. A. Taha, E. M. Akay, T. Kossen, J. Sobesky, J. D. Kelleher, K. Hildebrand, D. Frey, et al., A u-net deep learning framework for high performance vessel segmentation in patients with cerebrovascular disease, Frontiers in neuroscience 13 (2019) 97

    work page 2019

  22. [22]

    D. Chen, Y . Ao, S. Liu, Semi-supervised learning method of u-net deep learning network for blood vessel segmentation in retinal images, Sym- metry 12 (7) (2020) 1067

    work page 2020

  23. [23]

    Chatterjee, K

    S. Chatterjee, K. Prabhu, M. Pattadkal, G. Bortsova, C. Sarasaen, F. Dubost, H. Mattern, M. de Bruijne, O. Speck, A. N ¨urnberger, Ds6: Deformation-aware semi-supervised learning: Application to small vessel segmentation with noisy training data, Journal of Imaging 8 (10) (2022) 259

    work page 2022

  24. [24]

    Hanke, F

    M. Hanke, F. J. Baumgartner, P. Ibe, F. R. Kaule, S. Pollmann, O. Speck, W. Zinke, J. Stadler, A high-resolution 7-tesla fmri dataset from complex natural stimulation with an audio movie, Scientific data 1 (1) (2014) 1–18

    work page 2014

  25. [25]

    Fedorov, R

    A. Fedorov, R. Beichel, J. Kalpathy-Cramer, J. Finet, J.-C. Fillion-Robin, S. Pujol, C. Bauer, D. Jennings, F. Fennessy, M. Sonka, et al., 3d slicer as an image computing platform for the quantitative imaging network, Magnetic resonance imaging 30 (9) (2012) 1323–1341

    work page 2012

  26. [26]

    Chatterjee, H

    S. Chatterjee, H. Mattern, F. Dubost, S. Schreiber, A. N ¨urnberger, O. Speck, Smile-uhura: Small vessel segmentation at mesoscopic scale from ultra-high resolution 7t magnetic resonance angiograms (2022). doi:10.7303/SYN47164761. URL https://repo-prod.prod.sagebase.org/repo/v1/doi/ locate?id=syn47164761&type=ENTITY

    work page doi:10.7303/syn47164761 2022

  27. [27]

    Chatterjee, F

    S. Chatterjee, F. Gaidzik, A. Sciarra, H. Mattern, G. Janiga, O. Speck, A. N¨urnberger, S. Pathiraja, Pulaski: Learning inter-rater variability using statistical distances to improve probabilistic segmentation, arXiv preprint arXiv:2312.15686

    work page arXiv

  28. [28]

    Mattern, Openly available sMall vEsseL sEgmenTaTion pipelinE (OMELETTE), in: 2021 ISMRM & SMRT Annual Meeting & Exhibi- tion, 2021, p

    H. Mattern, Openly available sMall vEsseL sEgmenTaTion pipelinE (OMELETTE), in: 2021 ISMRM & SMRT Annual Meeting & Exhibi- tion, 2021, p. 3745

    work page 2021

  29. [29]

    S. Kohl, B. Romera-Paredes, C. Meyer, J. De Fauw, J. R. Ledsam, K. Maier-Hein, S. Eslami, D. Jimenez Rezende, O. Ronneberger, A prob- abilistic u-net for segmentation of ambiguous images, Advances in neural information processing systems 31

    work page

  30. [30]

    O. U. Aydin, A. A. Taha, A. Hilbert, A. A. Khalil, I. Galinovic, J. B. Fiebach, D. Frey, V . I. Madai, On the usage of average hausdorffdistance for segmentation performance assessment: hidden error when used for ranking, European radiology experimental 5 (2021) 1–7

    work page 2021

  31. [31]

    A. A. Taha, A. Hanbury, Metrics for evaluating 3d medical image seg- mentation: analysis, selection, and tool, BMC medical imaging 15 (2015) 1–28

    work page 2015

  32. [32]

    G. Zeng, X. Yang, J. Li, L. Yu, P.-A. Heng, G. Zheng, 3d u-net with multi-level deep supervision: fully automatic segmentation of proximal femur in 3d mr images, in: Machine Learning in Medical Imaging: 8th International Workshop, MLMI 2017, Held in Conjunction with MICCAI 2017, Quebec City, QC, Canada, September 10, 2017, Proceedings 8, Springer, 2017, p...

    work page 2017

  33. [33]

    X. Yu, Q. Yang, Y . Zhou, L. Y . Cai, R. Gao, H. H. Lee, T. Li, S. Bao, Z. Xu, T. A. Lasko, et al., Unest: local spatial representation learning with hierarchical transformer for efficient medical segmentation, Medical Image Analysis 90 (2023) 102939

    work page 2023

  34. [34]

    Isensee, P

    F. Isensee, P. F. Jaeger, S. A. Kohl, J. Petersen, K. H. Maier-Hein, nnu- net: a self-configuring method for deep learning-based biomedical image segmentation, Nature methods 18 (2) (2021) 203–211

    work page 2021

  35. [35]

    M. J. Cardoso, W. Li, R. Brown, N. Ma, E. Kerfoot, Y . Wang, B. Murrey, A. Myronenko, C. Zhao, D. Yang, et al., Monai: An open-source frame- work for deep learning in healthcare, arXiv preprint arXiv:2211.02701

    work page internal anchor Pith review Pith/arXiv arXiv

  36. [36]

    Loshchilov, F

    I. Loshchilov, F. Hutter, Decoupled weight decay regularization, in: In- ternational Conference on Learning Representations, 2019

    work page 2019

  37. [37]

    Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, B. Guo, Swin transformer: Hierarchical vision transformer using shifted windows, in: Proceedings of the IEEE /CVF international conference on computer vi- sion, 2021, pp. 10012–10022

    work page 2021

  38. [38]

    Hatamizadeh, V

    A. Hatamizadeh, V . Nath, Y . Tang, D. Yang, H. R. Roth, D. Xu, Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images, in: International MICCAI brainlesion workshop, Springer, 2021, pp. 272–284

    work page 2021

  39. [39]

    M. Xu, F. L. Ribeiro, M. Barth, M. Bernier, S. Bollmann, S. Chatterjee, F. Cognolato, O. F. Gulban, V . Itkyal, S. Liu, H. Mattern, J. R. Polimeni, T. B. Shaw, O. Speck, S. Bollmann, Vesselboost: A Python Toolbox for 34 Small Blood Vessel Segmentation in Human Magnetic Resonance An- giography Data, Aperture Neuro 4. doi:10.52294/001c.123217

    work page doi:10.52294/001c.123217

  40. [40]

    C ¸ ic ¸ek, A

    ¨O. C ¸ ic ¸ek, A. Abdulkadir, S. S. Lienkamp, T. Brox, O. Ronneberger, 3d u-net: learning dense volumetric segmentation from sparse annota- tion, in: Medical Image Computing and Computer-Assisted Intervention– MICCAI 2016: 19th International Conference, Athens, Greece, October 17-21, 2016, Proceedings, Part II 19, Springer, 2016, pp. 424–432

    work page 2016

  41. [41]

    H. Mattern, Openly available small vessel segmentation pipeline (omelette), 29th Annual Meeting of International Society of Magnetic Resonance in Medicine (ISMRM), May 2021, virtual meeting, 2021. URL https://archive.ismrm.org/2021/3745.html

    work page 2021

  42. [42]

    N. J. Tustison, B. B. Avants, P. A. Cook, Y . Zheng, A. Egan, P. A. Yushke- vich, J. C. Gee, N4itk: improved n3 bias correction, IEEE transactions on medical imaging 29 (6) (2010) 1310–1320

    work page 2010

  43. [43]

    J. V . Manj ´on, P. Coup ´e, L. Mart ´ı-Bonmat´ı, D. L. Collins, M. Robles, Adaptive non-local means denoising of mr images with spatially vary- ing noise levels, Journal of Magnetic Resonance Imaging 31 (1) (2010) 192–203

    work page 2010

  44. [44]

    Radhakrishna, K

    C. Radhakrishna, K. V . Chintalapati, S. C. H. R. Kumar, R. Sutrave, H. Mattern, O. Speck, A. N ¨urnberger, S. Chatterjee, Spockmip: Seg- mentation of vessels in mras with enhanced continuity using maximum intensity projection as loss, arXiv preprint arXiv:2407.08655

    work page arXiv

  45. [45]

    Abraham, N

    N. Abraham, N. M. Khan, A novel focal tversky loss function with im- proved attention u-net for lesion segmentation, in: 2019 IEEE 16th inter- national symposium on biomedical imaging (ISBI 2019), IEEE, 2019, pp. 683–687

    work page 2019

  46. [46]

    Valderrama, I

    N. Valderrama, I. Pitsiorlas, L. Vargas, P. Arbel ´aez, M. A. Zuluaga, Job- vs: Joint brain-vessel segmentation in tof-mra images, in: 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI), IEEE, 2023, pp. 1–5. 35

    work page 2023