pith. machine review for the scientific record. sign in

arxiv: 2605.02737 · v1 · submitted 2026-05-04 · 💻 cs.CV

Recognition: 2 theorem links

· Lean Theorem

SIAM: Head and Brain MRI Segmentation from Few High-Quality Templates via Synthetic Training

Romain Valabregue , Ines Khemir , Eric Badinet , Fran\c{c}ois Rousseau , Guillaume Auzias , Reuben Dorent
Authors on Pith no claims yet

Pith reviewed 2026-05-08 18:33 UTC · model grok-4.3

classification 💻 cs.CV
keywords MRI segmentationsynthetic datadomain randomizationwhole head segmentationbrain structuresextra-cerebral tissues3D segmentation
0
0 comments X

The pith

A segmentation model trained synthetically from only six manual templates matches or outperforms current methods for whole-head MRI across varied datasets.

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

The paper presents SIAM, a framework for segmenting 16 structures in 3D head scans using synthetic data generated from just six high-quality templates. It applies randomization to image intensities and high-resolution spatial changes to model differences in shape and contrast. This enables segmentation of brain areas plus extra-cerebral tissues like the skull and skin, without needing preprocessing steps. Evaluation on eight diverse datasets with 301 subjects shows it performs as well as or better than existing approaches for brain structures while adding capability for non-brain ones.

Core claim

SIAM shows that extending domain randomization to both intensity and shape domains allows training a robust 3D segmentation model from a minimal set of six manually annotated templates, achieving accurate whole-head segmentation including brain and non-brain tissues on heterogeneous data.

What carries the argument

Domain randomization via synthetic image generation for contrast variability and high-resolution spatial transformations for anatomical differences, applied to six templates to train the model.

Load-bearing premise

The limited set of six templates combined with randomization is enough to represent all real-world variations in anatomy and image contrast without systematic errors.

What would settle it

Performance significantly below state-of-the-art on a new dataset with unseen age groups or contrasts would indicate the assumption does not hold.

Figures

Figures reproduced from arXiv: 2605.02737 by Eric Badinet, Fran\c{c}ois Rousseau, Guillaume Auzias, Ines Khemir, Reuben Dorent, Romain Valabregue.

Figure 1
Figure 1. Figure 1: A) Synthetic data generation as originally proposed by Billot et al. B) Our approach. C) Importance of appropriate high￾resolution upsampling: panels C1 and C2 share the same underlying resolution. With nearest-neighbor interpolation (C1), the original 0.5-mm voxel grid remains apparent, with no effective gain in resolution. In contrast, our approach (C2) generates smooth boundaries between structures, rev… view at source ↗
Figure 2
Figure 2. Figure 2: Anatomical accuracy against reference annotations: (A) Cortical Gray Matter (GM) Dice scores across 7 datasets. (B) Subcortical Dice evaluated on MICCAI 2012 (manual reference) and Mindboggle (FreeSurfer reference). (C) GM and combined deep nuclei Dice scores on the DBB dataset, separating 4 subjects with severe hydrocephalus (XXL ventricles) from the total cohort. (D) Skull Dice evaluation on the private … view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative segmentation examples. (A, B) Putamen: all models and the FreeSurfer REF include part of the claustrum, whereas only SIAM and the MICCAI 2012 REF are anatomically correct. (C) Cerebellum: only SIAM captures finer GM/WM details and does not include veins. (D, E) Two DBB outliers with comparable Dice scores but distinct error: sources—reference error in (D) and prediction error in (E). 10 view at source ↗
Figure 4
Figure 4. Figure 4: (A) Prediction consistency: Dice evaluation on the HCP test set, comparing T1w versus T1w repeat and T1w versus T2w. (B–D) Sensitivity to GM atrophy: (B) average Dice scores for subjects with and without atrophy; (C) absolute predicted volumes (with reference volume marked as ×; (D) relative atrophy prediction errors as defined in 1. Despite similar average Dice scores, the accuracy of relative atrophy pre… view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative segmentation examples. (A) ULTRACORTEX examples, where the GOUHFI brain mask erodes part of the GM (yellow arrows). (B) HCP examples of vessel segmentation. T2w predictions are denser, as small vessels are more visible on this sequence. 5.3 Sensitivity to cortical atrophy To assess sensitivity to cortical changes, we utilized the SynthAtrophy dataset, consisting of T1w scans from 20 subjects wi… view at source ↗
read the original abstract

Synthetic training has recently advanced brain MRI segmentation by enabling contrast-agnostic models trained entirely on generated data. However, most existing approaches rely on hundreds of automatically labeled templates, introducing systematic biases and limiting their flexibility to incorporate new anatomical structures. We present the Segment It All Model (SIAM), a 3D whole-head segmentation framework for 16 anatomical structures, trained using only six high-quality, manually annotated templates. SIAM extends domain randomization to both intensity and shape domains: synthetic image generation ensures contrast variability, while high-resolution spatial transformations model anatomical differences in cortical thickness and deep nuclei morphology. Unlike prior synthetic models, SIAM simultaneously segments brain as well as extra-cerebral tissues, including cerebrospinal fluid, vessels, dura mater, skull, and skin, enabling fully automated, preprocessing-free analysis. Evaluation across eight heterogeneous datasets (N=301), that include multiple contrasts (T1-weighted, T2-weighted, CT) and span a wide range of ages, demonstrates that SIAM matches or outperforms state-of-the-art methods for brain structures, in addition to extending automated segmentation to non-brain structures. The model also exhibits superior consistency across contrasts and repeated acquisitions, together with improved sensitivity to subtle gray matter atrophy. We openly release the model and the label templates at https://github.com/romainVala/SIAM.

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

3 major / 2 minor

Summary. The paper introduces SIAM, a 3D whole-head segmentation model for 16 anatomical structures (brain plus extra-cerebral tissues such as CSF, vessels, dura, skull, and skin) trained exclusively on synthetic images generated from only six high-quality manually annotated templates. It extends domain randomization to both intensity and high-resolution spatial transformations to model contrast and anatomical variability, claims to match or exceed state-of-the-art performance on eight heterogeneous real-world datasets (N=301) spanning T1w, T2w, CT contrasts and wide age ranges, and reports improved cross-contrast consistency and sensitivity to subtle gray-matter atrophy. The model and templates are released openly.

Significance. If the central claims hold, the work is significant for reducing dependence on large manually labeled datasets in medical image segmentation, enabling preprocessing-free multi-structure head segmentation, and demonstrating that limited high-quality templates plus targeted randomization can generalize across contrasts and populations. The open release of the model and label templates strengthens reproducibility and potential impact.

major comments (3)
  1. [Methods] Methods (synthetic generation pipeline): the claim that high-resolution spatial transformations plus intensity randomization from six templates suffice to cover anatomical and contrast variability across the eight test datasets lacks supporting quantitative evidence such as feature-space coverage statistics, maximum mean discrepancy, or outlier analysis between the synthetic distribution and real test-set tails; without this, aggregate Dice/HD metrics on N=301 cannot rule out systematic under-representation of certain age or morphology regimes.
  2. [Results] Results (evaluation tables): while overall performance is reported to match or exceed SOTA, the manuscript provides no subgroup breakdown (e.g., by age quartile or contrast) or failure-case analysis that would confirm the six-template manifold reaches the tails of the heterogeneous test distributions; this is load-bearing for the generalization claim.
  3. [Discussion] Discussion (limitations): the absence of any ablation on template count or diversity (e.g., performance drop when using 3 vs. 6 templates) leaves open whether the reported gains are robust to the specific choice of the six templates or would degrade on unseen morphological variants.
minor comments (2)
  1. [Abstract] Abstract and introduction: the phrase 'matches or outperforms' should be accompanied by a precise statement of which structures and datasets show statistically significant gains versus which show parity.
  2. [Figures] Figure captions: several figures comparing synthetic vs. real images would benefit from explicit scale bars and intensity histograms to allow readers to assess the randomization range.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback and for recognizing the potential significance of SIAM in reducing reliance on large manually labeled datasets. We address each major comment below, providing our response and indicating planned revisions to the manuscript.

read point-by-point responses

  1. Referee: [Methods] Methods (synthetic generation pipeline): the claim that high-resolution spatial transformations plus intensity randomization from six templates suffice to cover anatomical and contrast variability across the eight test datasets lacks supporting quantitative evidence such as feature-space coverage statistics, maximum mean discrepancy, or outlier analysis between the synthetic distribution and real test-set tails; without this, aggregate Dice/HD metrics on N=301 cannot rule out systematic under-representation of certain age or morphology regimes.

    Authors: We agree that direct quantitative measures such as MMD or feature-space coverage statistics would provide stronger support for the coverage claim. The current manuscript relies on empirical results across 301 scans from eight heterogeneous datasets as indirect evidence of generalization. In the revision, we will add supplementary material with intensity histogram comparisons and basic morphological statistics (e.g., structure volume distributions) between the synthetic data generated from the six templates and the real test sets to better characterize coverage of variability. revision: partial

  2. Referee: [Results] Results (evaluation tables): while overall performance is reported to match or exceed SOTA, the manuscript provides no subgroup breakdown (e.g., by age quartile or contrast) or failure-case analysis that would confirm the six-template manifold reaches the tails of the heterogeneous test distributions; this is load-bearing for the generalization claim.

    Authors: We acknowledge that subgroup breakdowns and failure-case analysis would strengthen the evidence for reaching the tails of the distributions. We will revise the results section to include performance breakdowns by contrast type (T1w, T2w, CT) and by available age categories across the datasets. We will also add a discussion of challenging cases with qualitative examples to illustrate performance on potential outliers. revision: yes

  3. Referee: [Discussion] Discussion (limitations): the absence of any ablation on template count or diversity (e.g., performance drop when using 3 vs. 6 templates) leaves open whether the reported gains are robust to the specific choice of the six templates or would degrade on unseen morphological variants.

    Authors: We recognize that an ablation on template count would be valuable for assessing robustness. However, performing multiple full retraining experiments is computationally prohibitive. We will expand the limitations section to explicitly acknowledge the dependence on the chosen six high-quality templates and to suggest that future studies could investigate the effects of template number and diversity on generalization to unseen morphologies. revision: partial

Circularity Check

0 steps flagged

No significant circularity; empirical evaluation on independent real data

full rationale

The paper describes generating synthetic training images from six independent, manually annotated templates using high-resolution spatial transformations and intensity randomization, then training a 3D segmentation network and reporting empirical Dice/HD metrics on eight separate real-world datasets (N=301) spanning multiple contrasts and ages. Performance claims are direct comparisons against SOTA methods on held-out external data, with no parameter fitting to test distributions, no self-definitional reductions (e.g., no ratio or coverage metric defined from the same templates and then re-predicted), and no load-bearing self-citations or uniqueness theorems invoked. The central result remains falsifiable against real data and does not collapse to the synthetic generation procedure by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that synthetic data generated from a small number of templates via randomization will generalize to real heterogeneous scans. No free parameters or new invented entities are described in the abstract.

axioms (1)
  • domain assumption Domain randomization over intensity and high-resolution spatial transformations applied to a small set of templates produces training data whose distribution is close enough to real scans for effective segmentation learning.
    This assumption underpins the entire synthetic training strategy and is invoked to justify using only six templates instead of hundreds.

pith-pipeline@v0.9.0 · 5555 in / 1418 out tokens · 66653 ms · 2026-05-08T18:33:00.269710+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

187 extracted references · 72 canonical work pages · 1 internal anchor

  1. [2]

    and Hiess, Rebecca Kucharsky and Kuzniecky, Ruben , year =

    Pardoe, Heath R. and Hiess, Rebecca Kucharsky and Kuzniecky, Ruben , year =. Motion and morphometry in clinical and nonclinical populations , volume =. Neuroimage , publisher =

  2. [3]

    Dylan and Qureshi, Abid and Buckner, Randy L

    Reuter, Martin and Tisdall, M. Dylan and Qureshi, Abid and Buckner, Randy L. and van der Kouwe, André JW and Fischl, Bruce , year =. Head motion during. Neuroimage , publisher =

  3. [4]

    Magnetic resonance imaging and the reduction of motion artifacts: review of the principles , volume =

    Van de Walle, Rik and Lemahieu, Ignace and Achten, Eric , year =. Magnetic resonance imaging and the reduction of motion artifacts: review of the principles , volume =. Technology and Health Care , publisher =

  4. [5]

    Motion artifacts in

    Zaitsev, Maxim and Maclaren, Julian and Herbst, Michael , year =. Motion artifacts in. Journal of Magnetic Resonance Imaging , publisher =

  5. [6]

    Blind retrospective motion correction of

    Loktyushin, Alexander and Nickisch, Hannes and Pohmann, Rolf and Schölkopf, Bernhard , year =. Blind retrospective motion correction of. Magnetic resonance in medicine , publisher =

  6. [8]

    and Gruetter, Rolf , year =

    Gallichan, Daniel and Marques, José P. and Gruetter, Rolf , year =. Retrospective correction of involuntary microscopic head movement using highly accelerated fat image navigators (. Magnetic resonance in medicine , publisher =

  7. [9]

    Jon and Egan, Gary F

    Pawar, Kamlesh and Chen, Zhaolin and Shah, N. Jon and Egan, Gary F. , year =. Suppressing motion artefacts in. NMR in Biomedicine , publisher =

  8. [10]

    and Drangova, Maria , year =

    Johnson, Patricia M. and Drangova, Maria , year =. Conditional generative adversarial network for. Magnetic resonance in medicine , publisher =

  9. [11]

    and Varsavsky, Thomas and Ourselin, Sébastien and Cardoso, M

    Shaw, Richard and Sudre, Carole H. and Varsavsky, Thomas and Ourselin, Sébastien and Cardoso, M. Jorge , year =. A k-. IEEE transactions on medical imaging , publisher =

  10. [12]

    and Zhao, Lu and Sepehrband, Farshid and Min, Joyce and Wang, Danny JJ and Shi, Yonggang and Toga, Arthur W

    Duffy, Ben A. and Zhao, Lu and Sepehrband, Farshid and Min, Joyce and Wang, Danny JJ and Shi, Yonggang and Toga, Arthur W. and Kim, Hosung and Initiative, Alzheimer's Disease Neuroimaging , year =. Retrospective motion artifact correction of structural. NeuroImage , publisher =

  11. [13]

    Retrospective correction of rigid and non-rigid

    Armanious, Karim and Gatidis, Sergios and Nikolaou, Konstantin and Yang, Bin and Kustner, Thomas , year =. Retrospective correction of rigid and non-rigid. 2019

    2019

  12. [14]

    Automated reference-free detection of motion artifacts in magnetic resonance images , volume =

    Küstner, Thomas and Liebgott, Annika and Mauch, Lukas and Martirosian, Petros and Bamberg, Fabian and Nikolaou, Konstantin and Yang, Bin and Schick, Fritz and Gatidis, Sergios , year =. Automated reference-free detection of motion artifacts in magnetic resonance images , volume =. Magnetic Resonance Materials in Physics, Biology and Medicine , publisher =

  13. [15]

    and van der Zwaag, Wietske and Gallichan, Daniel and Alkemade, Anneke and Vos, Frans M

    Bazin, Pierre-Louis and Nijsse, Hannah E. and van der Zwaag, Wietske and Gallichan, Daniel and Alkemade, Anneke and Vos, Frans M. and Forstmann, Birte U. and Caan, Matthan WA , year =. Sharpness in motion corrected quantitative imaging at. NeuroImage , publisher =

  14. [16]

    Motion artifacts reduction in brain

    Liu, Junchi and Kocak, Mehmet and Supanich, Mark and Deng, Jie , year =. Motion artifacts reduction in brain. Magnetic Resonance Imaging , publisher =

  15. [17]

    Motion artifact reduction using a convolutional neural network for dynamic contrast enhanced

    Tamada, Daiki and Kromrey, Marie-Luise and Ichikawa, Shintaro and Onishi, Hiroshi and Motosugi, Utaroh , year =. Motion artifact reduction using a convolutional neural network for dynamic contrast enhanced. Magnetic Resonance in Medical Sciences , publisher =

  16. [18]

    Means and averaging in the group of rotations , volume =

    Moakher, Maher , year =. Means and averaging in the group of rotations , volume =. SIAM journal on matrix analysis and applications , publisher =

  17. [19]

    Gallichan, Daniel , month = may, year =

  18. [20]

    2021 , note =

    finufft , copyright =. 2021 , note =

    2021

  19. [21]

    NeuroImage , author =

    Image processing and. NeuroImage , author =. 2018 , keywords =. doi:10.1016/j.neuroimage.2017.10.034 , abstract =

    work page doi:10.1016/j.neuroimage.2017.10.034 2018

  20. [23]

    Unsupervised medical image translation using

    Armanious, Karim and Jiang, Chenming and Abdulatif, Sherif and Küstner, Thomas and Gatidis, Sergios and Yang, Bin , year =. Unsupervised medical image translation using. 2019 27th

    2019

  21. [24]

    , year =

    Hanson, Andrew J. , year =. The quaternion-based spatial-coordinate and orientation-frame alignment problems , volume =. Acta Crystallographica Section A: Foundations and Advances , publisher =

  22. [25]

    Linear combination of transformations , volume =

    Alexa, Marc , year =. Linear combination of transformations , volume =. ACM Transactions on Graphics (TOG) , publisher =

  23. [27]

    The dual quaternion approach to hand-eye calibration , volume =

    Daniilidis, Konstantinos and Bayro-Corrochano, Eduardo , year =. The dual quaternion approach to hand-eye calibration , volume =. proceedings of 13th

  24. [28]

    Dual quaternions for rigid transformation blending , journal =

    Kavan, Ladislav and Collins, Steven and O’Sullivan, Carol and Zara, Jiri , year =. Dual quaternions for rigid transformation blending , journal =

  25. [29]

    and Barnes, Kelly A

    Power, Jonathan D. and Barnes, Kelly A. and Snyder, Abraham Z. and Schlaggar, Bradley L. and Petersen, Steven E. , year =. Spurious but systematic correlations in functional connectivity. Neuroimage , publisher =

  26. [30]

    Studholme, C., Hill, DLG, Hawkes, DJ , author =

    Measuring transformation error by. Studholme, C., Hill, DLG, Hawkes, DJ , author =. 1999 , file =

    1999

  27. [31]

    and Lutti, Antoine and Weiskopf, Nikolaus , year =

    Todd, Nick and Josephs, Oliver and Callaghan, Martina F. and Lutti, Antoine and Weiskopf, Nikolaus , year =. Prospective motion correction of. NeuroImage , publisher =

  28. [32]

    IEEE Transactions on Medical Imaging , author =

    Toblerone:. IEEE Transactions on Medical Imaging , author =. 2020 , note =. doi:10.1109/TMI.2019.2951080 , abstract =

    work page doi:10.1109/tmi.2019.2951080 2020

  29. [33]

    and Sotiropoulos, Stamatios N

    Glasser, Matthew F. and Sotiropoulos, Stamatios N. and Wilson, J. Anthony and Coalson, Timothy S. and Fischl, Bruce and Andersson, Jesper L. and Xu, Junqian and Jbabdi, Saad and Webster, Matthew and Polimeni, Jonathan R. , year =. The minimal preprocessing pipelines for the. Neuroimage , publisher =

  30. [35]

    Improved optimization for the robust and accurate linear registration and motion correction of brain images , volume =

    Jenkinson, Mark and Bannister, Peter and Brady, Michael and Smith, Stephen , year =. Improved optimization for the robust and accurate linear registration and motion correction of brain images , volume =. Neuroimage , publisher =

  31. [36]

    , year =

    Huggins, David J. , year =. Comparing distance metrics for rotation using the k-nearest neighbors algorithm for entropy estimation , volume =. Journal of computational chemistry , publisher =

  32. [37]

    Homogeneous coordinates in motion correction , volume =

    Zahneisen, Benjamin and Ernst, Thomas , year =. Homogeneous coordinates in motion correction , volume =. Magnetic resonance in medicine , publisher =

  33. [38]

    and Reisert, Marco , year =

    Kellner, Elias and Dhital, Bibek and Kiselev, Valerij G. and Reisert, Marco , year =. Gibbs-ringing artifact removal based on local subvoxel-shifts , volume =. Magnetic resonance in medicine , publisher =

  34. [39]

    and Keevil, Steven F

    Atkinson, David and Hill, Derek LG and Stoyle, Peter NR and Summers, Paul E. and Keevil, Steven F. , year =. Automatic correction of motion artifacts in magnetic resonance images using an entropy focus criterion , volume =. IEEE Transactions on Medical imaging , publisher =

  35. [40]

    Cortical thickness and central surface estimation , volume =

    Dahnke, Robert and Yotter, Rachel Aine and Gaser, Christian , year =. Cortical thickness and central surface estimation , volume =. Neuroimage , publisher =

  36. [41]

    and Ackerman, Michael J

    Yoo, Terry S. and Ackerman, Michael J. and Lorensen, William E. and Schroeder, Will and Chalana, Vikram and Aylward, Stephen and Metaxas, Dimitris and Whitaker, Ross , year =. Engineering and algorithm design for an image processing. Medicine

  37. [43]

    and Li, Wenqi and Vercauteren, Tom and Ourselin, Sebastien and Jorge Cardoso, M

    Sudre, Carole H. and Li, Wenqi and Vercauteren, Tom and Ourselin, Sebastien and Jorge Cardoso, M. , year =. Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations , booktitle =

  38. [44]

    Computer Methods and Programs in Biomedicine , publisher =

    Pérez-García, Fernando and Sparks, Rachel and Ourselin, Sébastien , year =. Computer Methods and Programs in Biomedicine , publisher =

  39. [45]

    and Wyss, Michael , year =

    Iacono, Maria Ida and Neufeld, Esra and Akinnagbe, Esther and Bower, Kelsey and Wolf, Johanna and Vogiatzis Oikonomidis, Ioannis and Sharma, Deepika and Lloyd, Bryn and Wilm, Bertram J. and Wyss, Michael , year =. PloS one , publisher =

  40. [46]

    and Sanroma, Gerard and Zimmer, Veronika A

    Benkarim, Oualid M. and Sanroma, Gerard and Zimmer, Veronika A. and Muñoz-Moreno, Emma and Hahner, Nadine and Eixarch, Elisenda and Camara, Oscar and Gonzalez Ballester, Miguel Angel and Piella, Gemma , year =. Toward the automatic quantification of in utero brain development in. Human brain mapping , publisher =

  41. [47]

    and Ortinau, Cynthia M

    Dou, Haoran and Karimi, Davood and Rollins, Caitlin K. and Ortinau, Cynthia M. and Vasung, Lana and Velasco-Annis, Clemente and Ouaalam, Abdelhakim and Yang, Xin and Ni, Dong and Gholipour, Ali , year =. A deep attentive convolutional neural network for automatic cortical plate segmentation in fetal. IEEE transactions on medical imaging , publisher =

  42. [48]

    arXiv preprint arXiv:2204.02779 , author =

    A. arXiv preprint arXiv:2204.02779 , author =. 2022 , file =

    work page arXiv 2022

  43. [49]

    and Warfield, Simon K

    Gholipour, Ali and Akhondi-Asl, Alireza and Estroff, Judy A. and Warfield, Simon K. , year =. Multi-atlas multi-shape segmentation of fetal brain. NeuroImage , publisher =

  44. [50]

    and Velasco-Annis, Clemente and Ouaalam, Abdelhakim and Gholipour, Ali , year =

    Karimi, Davood and Rollins, Caitlin K. and Velasco-Annis, Clemente and Ouaalam, Abdelhakim and Gholipour, Ali , year =. Learning to segment fetal brain tissue from noisy annotations , journal =

  45. [51]

    A global optimisation method for robust affine registration of brain images , volume =

    Jenkinson, Mark and Smith, Stephen , year =. A global optimisation method for robust affine registration of brain images , volume =. Medical image analysis , publisher =

  46. [52]

    FMRIB Analysis Group of the University of Oxford , author =

    Non-linear registration, aka. FMRIB Analysis Group of the University of Oxford , author =. 2007 , note =

    2007

  47. [53]

    and Taylor, Zeike A

    Modat, Marc and Ridgway, Gerard R. and Taylor, Zeike A. and Lehmann, Manja and Barnes, Josephine and Hawkes, David J. and Fox, Nick C. and Ourselin, Sébastien , year =. Fast free-form deformation using graphics processing units , volume =. Computer methods and programs in biomedicine , publisher =

  48. [54]

    and Shit, Suprosanna and Iqbal, Asim and Khan, Romesa and Kottke, Raimund and Grehten, Patrice , year =

    Payette, Kelly and de Dumast, Priscille and Kebiri, Hamza and Ezhov, Ivan and Paetzold, Johannes C. and Shit, Suprosanna and Iqbal, Asim and Khan, Romesa and Kottke, Raimund and Grehten, Patrice , year =. An automatic multi-tissue human fetal brain segmentation benchmark using the fetal tissue annotation dataset , volume =. Scientific Data , publisher =

  49. [55]

    arXiv preprint arXiv:2204.03522 , author =

    Global displacement induced by rigid motion simulation during. arXiv preprint arXiv:2204.03522 , author =. 2022 , file =

    work page arXiv 2022

  50. [56]

    and Tusor, Nora and Steinweg, Johannes and Hughes, Emer and Grande, Lucilio Cordero and Price, Anthony and Hutter, Jana , year =

    Schuh, Andreas and Makropoulos, Antonios and Wright, Robert and Robinson, Emma C. and Tusor, Nora and Steinweg, Johannes and Hughes, Emer and Grande, Lucilio Cordero and Price, Anthony and Hutter, Jana , year =. A deformable model for the reconstruction of the neonatal cortex , booktitle =

  51. [57]

    and Sedmak, Goran and Judaš, M

    Kostović, I. and Sedmak, Goran and Judaš, M. , year =. Neural histology and neurogenesis of the human fetal and infant brain , volume =. Neuroimage , publisher =

  52. [58]

    and Kohl, Simon AA and Petersen, Jens and Maier-Hein, Klaus H

    Isensee, Fabian and Jaeger, Paul F. and Kohl, Simon AA and Petersen, Jens and Maier-Hein, Klaus H. , year =. Nature methods , publisher =

  53. [59]

    and Rueckert, Daniel , year =

    Makropoulos, Antonios and Counsell, Serena J. and Rueckert, Daniel , year =. A review on automatic fetal and neonatal brain. NeuroImage , publisher =

  54. [60]

    and Schuh, Andreas and Wright, Robert and Fitzgibbon, Sean and Bozek, Jelena and Counsell, Serena J

    Makropoulos, Antonios and Robinson, Emma C. and Schuh, Andreas and Wright, Robert and Fitzgibbon, Sean and Bozek, Jelena and Counsell, Serena J. and Steinweg, Johannes and Vecchiato, Katy and Passerat-Palmbach, Jonathan , year =. The developing human connectome project:. Neuroimage , publisher =

  55. [61]

    and Ledig, Christian and Aljabar, Paul and Serag, Ahmed and Hajnal, Joseph V

    Makropoulos, Antonios and Gousias, Ioannis S. and Ledig, Christian and Aljabar, Paul and Serag, Ahmed and Hajnal, Joseph V. and Edwards, A. David and Counsell, Serena J. and Rueckert, Daniel , year =. Automatic whole brain. IEEE transactions on medical imaging , publisher =

  56. [62]

    NeuroImage , author =

    Computational neuroanatomy of baby brains:. NeuroImage , author =. 2019 , keywords =. doi:10.1016/j.neuroimage.2018.03.042 , abstract =

    work page doi:10.1016/j.neuroimage.2018.03.042 2019

  57. [63]

    A lifelong learning approach to brain

    Karani, Neerav and Chaitanya, Krishna and Baumgartner, Christian and Konukoglu, Ender , year =. A lifelong learning approach to brain. International

  58. [64]

    Magnetic Resonance in Medicine , author =

    A dedicated neonatal brain imaging system , volume =. Magnetic Resonance in Medicine , author =. 2017 , note =. doi:10.1002/mrm.26462 , abstract =

    work page doi:10.1002/mrm.26462 2017

  59. [65]

    Ronneberger, Olaf and Fischer, Philipp and Brox, Thomas , year =. U-. Medical. doi:10.1007/978-3-319-24574-4_28 , abstract =

    work page doi:10.1007/978-3-319-24574-4_28

  60. [66]

    and Hainfellner, Johannes and Bettelheim, Dieter and Langs, Georg , year =

    Pogledic, Ivana and Schwartz, Ernst and Mitter, Christian and Baltzer, Pascal and Milos, Ruxandra-Iulia and Gruber, Gerlinde Maria and Brugger, Peter C. and Hainfellner, Johannes and Bettelheim, Dieter and Langs, Georg , year =. The subplate layers: the superficial and deep subplate can be discriminated on 3 tesla human fetal postmortem. Cerebral Cortex ,...

  61. [67]

    and Dalca, Adrian V

    Zhao, Amy and Balakrishnan, Guha and Durand, Fredo and Guttag, John V. and Dalca, Adrian V. , year =. Data augmentation using learned transformations for one-shot medical image segmentation , booktitle =

  62. [68]

    Self-supervised generative style transfer for one-shot medical image segmentation , booktitle =

    Tomar, Devavrat and Bozorgtabar, Behzad and Lortkipanidze, Manana and Vray, Guillaume and Rad, Mohammad Saeed and Thiran, Jean-Philippe , year =. Self-supervised generative style transfer for one-shot medical image segmentation , booktitle =

  63. [69]

    Test-time adaptable neural networks for robust medical image segmentation , volume =

    Karani, Neerav and Erdil, Ertunc and Chaitanya, Krishna and Konukoglu, Ender , year =. Test-time adaptable neural networks for robust medical image segmentation , volume =. Medical Image Analysis , publisher =

  64. [70]

    2023 , keywords =

    Medical Image Analysis , author =. 2023 , keywords =. doi:10.1016/j.media.2023.102789 , abstract =

    work page doi:10.1016/j.media.2023.102789 2023

  65. [71]

    NeuroImage , author =

    An automated framework for localization, segmentation and super-resolution reconstruction of fetal brain. NeuroImage , author =. 2020 , keywords =. doi:10.1016/j.neuroimage.2019.116324 , abstract =

    work page doi:10.1016/j.neuroimage.2019.116324 2020

  66. [72]

    Jog, Amod and Fischl, Bruce , month = jul, year =. Pulse

  67. [73]

    and Vineet, Vibhav and Roth, Stefan and Koltun, Vladlen , month = aug, year =

    Richter, Stephan R. and Vineet, Vibhav and Roth, Stefan and Koltun, Vladlen , month = aug, year =. Playing for

  68. [74]

    Image-to-Image Translation with Conditional Adversarial Networks

    Isola, Phillip and Zhu, Jun-Yan and Zhou, Tinghui and Efros, Alexei A. , month = jul, year =. Image-to-. 2017. doi:10.1109/CVPR.2017.632 , abstract =

    work page doi:10.1109/cvpr.2017.632 2017

  69. [75]

    IEEE Transactions on Knowledge and Data Engineering 22, 1345–1359

    A. IEEE Transactions on Knowledge and Data Engineering , author =. 2010 , note =. doi:10.1109/TKDE.2009.191 , abstract =

    work page doi:10.1109/tkde.2009.191 2010

  70. [76]

    IEEE Transactions on Medical Imaging , author =

    Generalizing. IEEE Transactions on Medical Imaging , author =. 2020 , note =. doi:10.1109/TMI.2020.2973595 , language =

    work page doi:10.1109/tmi.2020.2973595 2020

  71. [77]

    Ganin, Yaroslav and Ustinova, Evgeniya and Ajakan, Hana and Germain, Pascal and Larochelle, Hugo and Laviolette, François and Marchand, Mario and Lempitsky, Victor , editor =. Domain-. Domain. 2017 , note =. doi:10.1007/978-3-319-58347-1_10 , abstract =

    work page doi:10.1007/978-3-319-58347-1_10 2017

  72. [78]

    Unsupervised domain adaptation in brain lesion segmentation with adversarial networks , booktitle =

    Kamnitsas, Konstantinos and Baumgartner, Christian and Ledig, Christian and Newcombe, Virginia and Simpson, Joanna and Kane, Andrew and Menon, David and Nori, Aditya and Criminisi, Antonio and Rueckert, Daniel , year =. Unsupervised domain adaptation in brain lesion segmentation with adversarial networks , booktitle =

  73. [79]

    Translating and segmenting multimodal medical volumes with cycle-and shape-consistency generative adversarial network , booktitle =

    Zhang, Zizhao and Yang, Lin and Zheng, Yefeng , year =. Translating and segmenting multimodal medical volumes with cycle-and shape-consistency generative adversarial network , booktitle =

  74. [80]

    Synergistic image and feature adaptation:

    Chen, Cheng and Dou, Qi and Chen, Hao and Qin, Jing and Heng, Pheng-Ann , year =. Synergistic image and feature adaptation:. Proceedings of the

  75. [81]

    IEEE Transactions on Medical Imaging , author =

    Multimodal. IEEE Transactions on Medical Imaging , author =. 2018 , note =. doi:10.1109/TMI.2017.2764326 , abstract =

    work page doi:10.1109/tmi.2017.2764326 2018

  76. [82]

    Domain randomization for transferring deep neural networks from simulation to the real world , booktitle =

    Tobin, Josh and Fong, Rachel and Ray, Alex and Schneider, Jonas and Zaremba, Wojciech and Abbeel, Pieter , year =. Domain randomization for transferring deep neural networks from simulation to the real world , booktitle =

  77. [83]

    Training deep networks with synthetic data:

    Tremblay, Jonathan and Prakash, Aayush and Acuna, David and Brophy, Mark and Jampani, Varun and Anil, Cem and To, Thang and Cameracci, Eric and Boochoon, Shaad and Birchfield, Stan , year =. Training deep networks with synthetic data:. Proceedings of the

  78. [84]

    Noise and the reality gap:

    Jakobi, Nick and Husbands, Phil and Harvey, Inman , year =. Noise and the reality gap:. Advances in

  79. [85]

    Journal of the Royal Statistical Society: Series B (Methodological) , author =

    Maximum. Journal of the Royal Statistical Society: Series B (Methodological) , author =. 1977 , note =. doi:10.1111/j.2517-6161.1977.tb01600.x , abstract =

    work page doi:10.1111/j.2517-6161.1977.tb01600.x 1977

  80. [86]

    Sokooti, Hessam and de Vos, Bob and Berendsen, Floris and Lelieveldt, Boudewijn P. F. and Išgum, Ivana and Staring, Marius , editor =. Nonrigid. Medical. 2017 , keywords =. doi:10.1007/978-3-319-66182-7_27 , abstract =

    work page doi:10.1007/978-3-319-66182-7_27 2017

Showing first 80 references.