Unified 3D MRI Representations via Sequence-Invariant Contrastive Learning

Cathy J. Price; Jenny Crinion; John Ashburner; Liam Chalcroft

REVIEW 4 major objections 4 minor 30 references

Unified 3D MRI Representations via Sequence-Invariant Contrastive Learning

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims that enforcing consistent representations across simulated MRI contrasts yields a single 3D encoder that learns anatomy rather than scan sequence, improving low-data segmentation and denoising.

desk verdict A practical new SSL idea for 3D MRI with a real reproducibility hole: the pretraining data provenance is never disclosed, so the headline gains are uninterpretable until that is fixed. read the letter →

arxiv 2501.12057 v3 pith:PHH63SGO submitted 2025-01-21 cs.CV physics.med-ph

Liam Chalcroft , Jenny Crinion , Cathy J. Price , John Ashburner This is my paper · ORCID

classification cs.CVphysics.med-ph

keywords self-supervisedlearningcontrastivequantitativeMRIsequence-invariantrepresentations3Dmedicalimagingbrainsegmentationdenoisingdomaingeneralisation

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

This paper tries to establish that forcing a self-supervised 3D encoder to agree across multiple synthetic MRI contrasts from the same underlying anatomy yields representations that are anatomy-centric rather than sequence-specific. The authors simulate T1-weighted, T2-weighted, PD-weighted, and FLAIR-like contrasts from quantitative MRI (qMRI) parameter maps, then train the encoder so that views from different simulated sequences produce consistent features. They report that the resulting single encoder improves healthy-brain segmentation, stroke-lesion segmentation, and denoising over a baseline pretrained on synthetic MPRAGE only, with the largest gains at 1% labelled data (up to +8.3 Dice points and +4.2 dB PSNR) and on unseen sites. If the claim holds, it offers a path to one protocol-agnostic backbone for clinical brain MRI that needs very little annotated data.

What carries the argument

The load-bearing mechanism is a physics-driven forward model paired with a sequence-invariance contrastive objective. Quantitative MRI parameter maps (proton density, R1, R2/R2*, and optionally magnetisation transfer) are passed through Bloch-equation signal equations for FSE, GRE, FLAIR, and MPRAGE with randomly sampled acquisition parameters and Rician noise, producing synthetic volumes that share anatomy but differ in contrast. The SeqInv variant then applies a contrastive loss that pulls together the latent representations of two different simulated sequences from the same volume, formally teaching the encoder to ignore which sequence produced the image; a reconstruction branch adds L1 pixel-level fidelity. SeqAug, which draws two views from the same simulated sequence, and Base, which uses only MPRAGE, serve as controls that isolate the effect of cross-sequence invariance.

What would settle it

Acquire paired qMRI and conventional T1w/T2w/PDw/FLAIR scans from the same subjects, pretrain the same three encoders on the qMRI maps, and compare Dice on the real conventional scans; if SeqInv's advantage over Base disappears on real data, the synthetic-to-real transfer claim fails.

Watch

Extended reading notes

Core claim

The central claim is that by simulating multiple MRI contrasts from a single 3D qMRI scan and enforcing consistent representations across these contrasts, sequence-invariant self-supervised learning captures anatomy rather than scanner-specific appearance. In the paper's comparison, the SeqInv encoder, whose two contrastive views come from different simulated sequences, outperforms the Base encoder, whose views come from synthetic MPRAGE only, across T1w, T2w, PDw, and FLAIR tasks on IXI and ARC data. The headline results are up to +8.3% Dice in low-label healthy-brain segmentation and +4.2 dB PSNR in denoising with 1% training data, with cross-site gains on the HH and IOP sites of IXI often exceeding in-domain gains. The authors also report partial sequence-specific gaps, particularly on T2w, indicating the learned invariance is not complete.

Load-bearing premise

The whole transfer argument rests on the forward signal equations and sampled acquisition parameters faithfully representing real scanner contrast, so invariance learned on synthetic pairs carries over to real T1w, T2w, PDw, and FLAIR scans.

Editorial extensions

If this is right

A single frozen 3D encoder can replace per-sequence and per-site models for brain MRI, since the same backbone transfers to T1w, T2w, PDw, and FLAIR across segmentation and denoising tasks.
Low-data clinical settings benefit most: at 1% of training labels the method reports up to +8.3 Dice points in segmentation and +4.2 dB in denoising over the baseline.
Cross-site deployment improves without any target-domain data, with gains on unseen sites often larger than gains in the training site.
Pretraining requires only qMRI parameter maps, which can be generated or estimated from existing structural MRI databases, making larger-scale pretraining a direct next step.
Sequence invariance is partial rather than total; the reported T2w gaps mean some sequence-specific cues are lost, which may matter when those cues are diagnostically useful.

Reading between the lines

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

The same sequence-invariance principle could extend to any modality with a forward signal model, such as CT from tissue maps or ultrasound from acoustic models, although the paper tests only MRI.
An encoder forced to preserve anatomy could serve as a generic feature extractor for registration, atlas building, or anomaly detection, not just the segmentation and denoising tasks evaluated here.
A paired dataset of qMRI and conventional scans from the same subjects would let a reader check directly whether the learned invariance carries over to real sequences, a test not reported in the paper.
Because pretraining used only 51 subjects, the reported gains may understate the method's ceiling if pretraining scales to larger qMRI cohorts, which the paper identifies as future work.

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

Two things you should know about arXiv:2501.12057. First, the core idea is genuinely new: train a 3D MRI encoder with SimCLR-style contrastive learning where the two views come from two different simulated MRI sequences, generated from qMRI parameter maps. The paper ships code, trained models, and the forward equations, so it's reproducible. Second, the paper has a hole that prevents me from trusting the headline numbers right now: it never says where the 51 qMRI pretraining volumes come from. The evaluation cohorts are IXI (healthy) and ARC (stroke); the pretraining set is 22 healthy and 29 stroke subjects. If those healthy or stroke subjects overlap with the test sets, the reported low-data and cross-site gains could be explained by subject leakage rather than sequence-invariant learning. The authors flag the small pretraining size as a limitation but do not address provenance.

On the positive side, the experimental design is thoughtful. Three models (Base, SeqAug, SeqInv) are compared under the same pretraining setup, so the contrast is clean. The low-data segmentation and denoising results in Tables 1 and 3 show a consistent direction: SeqInv helps, especially at 1% labels and out-of-domain sites. The discussion of why T2w is a weak spot is honest.

Soft spots, in rough order of importance. One, the missing provenance as above. Two, there is no significance testing; the word "significant" in the abstract is doing work that the paper doesn't back up with intervals or tests. Three, there are no real-MRI SSL baselines (e.g., MoCo or BYOL trained on actual T1w volumes), so we don't know how much of the gain is from physics-based synthesis versus the contrastive scheme. Four, results at 100% training data are mixed: SeqInv is not consistently better than Base, which the paper acknowledges but doesn't resolve. The synthetic-to-real transfer is plausible, and the out-of-domain IXI results give some evidence it works, but there is no paired synthetic-real sanity check.

Who should read this: anyone working on acquisition-invariant representations or low-data medical imaging. The method is practical and the code is there. My recommendation for peer review: send it to referees, but require the authors to disclose the exact provenance of the 51 qMRI volumes and show non-overlap with IXI and ARC before any acceptance. The contribution is worth referee time; the current version is not citable as evidence.

Referee Report

4 major / 4 minor

Summary. The paper proposes a self-supervised learning framework for 3D MRI that uses quantitative MRI (qMRI) parameter maps to simulate multiple MRI contrasts (FSE, GRE, FLAIR, MPRAGE) and trains an encoder to produce invariant representations across these sequences. The core idea is that enforcing consistency between distinct simulated contrasts yields anatomy-centric features rather than sequence-specific ones. The authors compare three pretraining strategies: Base (synthetic MPRAGE views), SeqAug (two views from the same simulated sequence), and SeqInv (two views from distinct simulated sequences). They evaluate on healthy brain segmentation (IXI), stroke lesion segmentation (ARC), and MRI denoising (IXI), reporting that SeqInv improves Dice and PSNR, especially with 1% labels and on out-of-domain sites, with headline gains of up to +8.3 Dice and +4.2 dB PSNR. They also release code and pretrained models.

Significance. If the results hold, the paper makes a valuable contribution by showing that physics-based simulation of diverse MRI contrasts can improve self-supervised representation learning for 3D medical imaging, with large gains in low-label and cross-site settings. The internal ablation (Base vs. SeqAug vs. SeqInv) cleanly isolates the effect of cross-sequence views, and the code release supports reproducibility. However, several load-bearing concerns must be addressed before the claims can be trusted: the provenance of the 51 pretraining qMRI volumes is undisclosed, statistical significance is not assessed, and the comparison set is limited to the authors' own baseline. These issues currently weaken the strength of the reported gains and the generality of the conclusions.

major comments (4)

[§4.1 and §5.2] The pretraining dataset is described only as "51 qMRI volumes (22 healthy, 29 stroke subjects)" without any statement of where these volumes come from. This is critical because the downstream benchmarks are IXI (a healthy multi-site dataset) and ARC (a chronic stroke repository): if any of the 51 volumes were drawn from IXI or ARC and overlap with the test subjects, the reported low-data and cross-site gains (e.g., +8.3 Dice, +4.2 dB PSNR) could be inflated by subject-level leakage rather than sequence-invariant learning. The authors must disclose the exact source of the qMRI volumes, confirm that no pretraining subject appears in any downstream train/validation/test split (e.g., by subject IDs), and update the limitations section accordingly. This is a load-bearing point for the central claim.
[Tables 1-3 and §5.1] The paper reports only means with standard errors and no significance testing, yet the abstract claims "significant gains." Several key comparisons have overlapping standard errors: for example, in Table 2 the ARC T1w DSC is 78.4±2.0 (Base) versus 78.9±1.9 (SeqInv), and the ARC T2w DSC for SeqAug (80.3±1.4) versus SeqInv (79.4±1.6) also overlaps. To support the claim of significance, the authors should provide paired tests (e.g., bootstrap or permutation over subjects) for the headline low-data and out-of-domain cells, or temper the language to describe the observed gains without the term "significant."
[§4.1 and abstract] The experimental comparison is limited to the authors' own baseline (SimCLR with a reconstruction branch, pretrained on synthetic MPRAGE). The abstract and introduction claim gains "over baseline SSL approaches," but no representative existing 3D medical SSL methods (e.g., Models Genesis [30] or Swin UNETR [24], both cited) are evaluated. Adding at least one standard SSL baseline would substantiate the claim that the sequence-invariant formulation offers benefits beyond the particular SimCLR setup used here.
[§3.2, Tables 4 and 5] The forward signal equations in Table 4 are simplified approximations (e.g., FSE ignores echo train length and refocusing pulses; FLAIR is a single-inversion model; MPRAGE uses a steady-state approximation), and the sampling ranges in Table 5 are very broad (e.g., FLAIR TR up to 5 s, FSE TR up to 3 s). The paper does not provide any paired synthetic-real validation that these simulations produce contrast variation representative of real T1w/T2w/PDw/FLAIR acquisitions. Since the method's mechanism is that invariance learned on synthetic sequences transfers to real sequences, the authors should include quantitative evidence (e.g., distribution comparisons or an ablation using real multi-contrast data) that the simulated contrast space covers the real imaging domain.

minor comments (4)

[§4.1 vs §3.1] Section 4.1 says the baseline is "real MPRAGE only," but Section 3.1 states that Base was "pretrained exclusively using synthetic MPRAGE images generated from qMRI parameter maps." This inconsistency should be corrected to avoid confusion about the experimental setup.
[§4.3] The sentence "looking at the individual tissue class metrics in Table 1" is inaccurate because Table 1 reports only a pooled Dice value per condition, not per-class metrics for grey matter, white matter, and CSF.
[Throughout] There are several typos and formatting issues: "prosed" should be "proposed" in Section 1; "96 3 patches" should be "96^3 patches" in Section 4.1; and the table headers contain "T raining Data" instead of "Training Data."
[Table 5] The units and rationale for some sampling ranges are unclear: the MPRAGE TR is listed as N(23, 2.3) without specifying units (likely seconds), and the FSE TR upper bound of 3 s is short for typical clinical FSE (often 3-6 s). Clarifying these choices would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the contrastive pretraining is defined from fixed physics simulations and external IXI/ARC benchmarks, with no downstream-fitted parameters.

full rationale

The paper's derivation chain is self-contained rather than circular. The pretraining objective combines a standard NT-Xent contrastive loss with an L1 reconstruction loss, and sequence-invariance is implemented by sampling two distinct simulated MRI sequences from fixed Bloch-equation forward models (Table 4) with pre-specified acquisition-parameter ranges (Table 5). The simulation parameters are stated before the experiments and are not fitted to any downstream metric. Downstream evaluation is performed with frozen pretrained encoders on external datasets (IXI for healthy-brain segmentation and denoising, ARC for stroke lesion segmentation), so the reported Dice and PSNR gains are not forced by construction. The self-citations in the Related Work (e.g., ref. [7], the authors' prior stroke-lesion synthesis paper) are contextual and not load-bearing for the main claim; no uniqueness theorem or prior result from the same authors is invoked to forbid alternative explanations. The only substantive concern is that the provenance of the 51 qMRI pretraining volumes is undisclosed, meaning a hidden overlap with IXI or ARC evaluation subjects cannot be formally excluded. That is a data-leakage and reproducibility risk, not a circularity of the derivation: nothing in the paper's equations or fitting procedure makes the downstream results equal to its inputs by definition. Therefore, under the stated rules requiring a quoted reduction or fitted-parameter-as-prediction, no circular step is established.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The method introduces no new physical entities. Its free parameters are the hand-chosen simulation ranges and noise levels that shape the synthetic training distribution. The key domain assumptions are that the physics-based synthesis is realistic and that the pretraining cohort is independent of the evaluation cohorts.

free parameters (2)

Simulated acquisition parameter sampling ranges = Ranges in Table 5, e.g., FLAIR TE logU(0.02, 0.10), MPRAGE TR N(23, 2.3)
Hand-chosen augmentation ranges for synthetic sequence generation. They determine the contrast diversity seen by the encoder but are not fitted to downstream metrics in the paper.
Rician noise level sigma = Not specified numerically in the main text; applied on-the-fly during synthesis
The noise level affects both the reconstruction task and the contrastive difficulty. No ablation is reported to show how sensitive the results are to this choice.

assumptions (3)

domain assumption The Bloch-equation forward models in Table 4 approximate real FSE, GRE, FLAIR, and MPRAGE signal formation across scanners.
The method relies on synthetic contrasts being realistic enough that invariance learned on them transfers to real T1w, T2w, PDw, and FLAIR images. No paired synthetic-real validation is shown. Invoked in Section 3.2 and Appendix A.
domain assumption Shared tissue parameter maps (PD, R1, R2*, MT) capture the anatomical information common to different MRI contrasts.
This is the core premise of qMRI-based sequence invariance, stated in Sections 1 and 3.2. If the maps do not capture the anatomy relevant for downstream tasks, the invariance target is ill-posed.
domain assumption The 51 pretraining qMRI volumes are representative of, and independent from, the downstream evaluation cohorts IXI and ARC.
The paper does not state the provenance of these volumes. The cross-site generalization claim depends on the encoder not having seen subjects from the test sites. See Section 4.1.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Unified 3D MRI Representations via Sequence-Invariant Contrastive Learning." pith.science (2026). https://pith.science/paper/PHH63SGO

@misc{pith2026250112057,
  author       = {Pith},
  title        = {Pith review of: Unified 3D MRI Representations via Sequence-Invariant Contrastive Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PHH63SGO}},
  note         = {Machine review of arXiv:2501.12057}
}

read the original abstract

Self-supervised deep learning has accelerated 2D natural image analysis but remains difficult to translate into 3D MRI, where data are scarce and pre-trained 2D backbones cannot capture volumetric context. We present a \emph{sequence-invariant} self-supervised framework leveraging quantitative MRI (qMRI). By simulating multiple MRI contrasts from a single 3D qMRI scan and enforcing consistent representations across these contrasts, we learn anatomy-centric rather than sequence-specific features. The result is a single 3D encoder that excels across tasks and protocols. Experiments on healthy brain segmentation (IXI), stroke lesion segmentation (ARC), and MRI denoising show significant gains over baseline SSL approaches, especially in low-data settings (up to +8.3\% Dice, +4.2 dB PSNR). It also generalises to unseen sites, supporting scalable clinical use. Code and trained models are publicly available at https://github.com/liamchalcroft/contrast-squared

Figures

Figures reproduced from arXiv: 2501.12057 by the authors.

**Figure 1.** Overview of the proposed SSL approach. (??) Baseline: An MPRAGE volume is augmented into two random views. We extract a feature vector h via the backbone encoder, project it to z for a contrastive loss Lcontrastive, and use a decoder to optimise a reconstruction/inpainting loss Lrecon. (??) SeqAug/SeqInv: We generalise this by simulating multiple scanner sequences from qMRI parameter maps, enabling sequence-invari… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 12 canonical work pages

[30]

Medical Image Analysis 67, 101840 (Jan 2021)

Zhou, Z., Sodha, V., Pang, J., Gotway, M.B., Liang, J.: Models genesis. Medical Image Analysis 67, 101840 (Jan 2021). https://doi.org/10.1016/j.media.2020. 101840, http://dx.doi.org/10.1016/j.media.2020.101840 12 Chalcroft et al. A Physics-based Signal Equations For each voxel we assume proton density (PD), longitudinal relaxation rate R1, transverse rela...

arXiv 2021
[24]

Tang, Y., Yang, D., Li, W., et al.: Self-supervised pre-training of swin transformers for 3d medical image analysis (2022), https://arxiv.org/abs/2111.14791 SeqInv: Sequence-Invariant Contrastive Learning for 3D MRI 11

arXiv 2022
[1]

https://doi.org/10.1101/2024.12.05.627056, http://dx.doi

Adams, R., Zhao, W., Hu, S., et al.: Ultimatesynth: Mri physics for pan-contrast ai (Dec 2024). https://doi.org/10.1101/2024.12.05.627056, http://dx.doi. org/10.1101/2024.12.05.627056

work page doi:10.1101/2024.12.05.627056 2024
[2]

Asiedu, E.B., Kornblith, S., Chen, T., et al.: Decoder denoising pretraining for semantic segmentation (2022), https://arxiv.org/abs/2205.11423

work page Pith review arXiv 2022
[3]

IEEE Transactions on Medi- cal Imaging 38(8), 1788–1800 (Aug 2019)

Balakrishnan, G., Zhao, A., Sabuncu, M.R., et al.: Voxelmorph: A learning frame- work for deformable medical image registration. IEEE Transactions on Medi- cal Imaging 38(8), 1788–1800 (Aug 2019). https://doi.org/10.1109/tmi.2019. 2897538, http://dx.doi.org/10.1109/TMI.2019.2897538

arXiv 2019
[4]

Medical Image Analysis 86, 102789 (May 2023)

Billot, B., Greve, D.N., Puonti, O., et al.: SynthSeg: Segmentation of brain MRI scans of any contrast and resolution without retraining. Medical Image Analysis 86, 102789 (May 2023). https://doi.org/10.1016/j.media.2023.102789

arXiv 2023
[5]

Borges, P., Shaw, R., Varsavsky, T., et al.: Acquisition-invariant brain mri segmen- tation with informative uncertainties (2021)

work page 2021
[6]

Medical Image Analysis 92, 103058 (Feb 2024)

Borges, P., Shaw, R., Varsavsky, T., et al.: Acquisition-invariant brain mri seg- mentation with informative uncertainties. Medical Image Analysis 92, 103058 (Feb 2024). https://doi.org/10.1016/j.media.2023.103058, http://dx.doi. org/10.1016/j.media.2023.103058

arXiv 2024

Show all 30 references

[7]

org/abs/2412.03318

Chalcroft, L., Crinion, J., Price, C.J., Ashburner, J.: Domain-agnostic stroke lesion segmentation using physics-constrained synthetic data (2024), https://arxiv. org/abs/2412.03318

2024 arXiv
[8]

Chalcroft, L., Pereira, R.L., Brudfors, M., et al.: Large-kernel attention for efficient and robust brain lesion segmentation (2023),https://arxiv.org/abs/2308.07251 10 Chalcroft et al

2023 arXiv
[9]

Chen, T., Kornblith, S., Norouzi, M., Hinton, G.: A simple framework for con- trastive learning of visual representations (2020), https://arxiv.org/abs/2002. 05709

2020
[10]

Medical Image Analysis 83, 102628 (Jan 2023)

Dorent, R., Kujawa, A., Ivory, M., et al.: Crossmoda 2021 challenge: Bench- mark of cross-modality domain adaptation techniques for vestibular schwan- noma and cochlea segmentation. Medical Image Analysis 83, 102628 (Jan 2023). https://doi.org/10.1016/j.media.2022.102628, http...

2021
[11]

Scientific Data 11(1), 1–8 (2024)

Gibson, M., Newman-Norlund, R., Bonilha, L., et al.: The Aphasia Recovery Co- hort, an open-source chronic stroke repository. Scientific Data 11(1), 1–8 (2024). https://doi.org/10.1038/s41597-024-03819-7 , http://dx.doi.org/10.1038/ s41597-024-03819-7

2024 doi
[12]

Grill, J.B., Strub, F., Altch´ e, F., et al.: Bootstrap your own latent: A new approach to self-supervised learning (2020), https://arxiv.org/abs/2006.07733

2020 arXiv
[13]

Magnetic Resonance in Medicine 34(6), 910–914 (Dec 1995)

Gudbjartsson, H., Patz, S.: The rician distribution of noisy mri data. Magnetic Resonance in Medicine 34(6), 910–914 (Dec 1995). https://doi.org/10.1002/ mrm.1910340618, http://dx.doi.org/10.1002/mrm.1910340618

1995 doi
[14]

org/abs/2506.04173

Hays, S.P., Zuo, L., Feng, A., et al.: Synthetic multi-inversion time magnetic res- onance images for visualization of subcortical structures (2025), https://arxiv. org/abs/2506.04173

2025
[15]

He, K., Fan, H., Wu, Y., et al.: Momentum contrast for unsupervised visual rep- resentation learning (2020), https://arxiv.org/abs/1911.05722

2020 arXiv
[16]

Nature 495(7440), 187–192 (Mar 2013)

Ma, D., Gulani, V., Seiberlich, N., et al.: Magnetic resonance fingerprinting. Nature 495(7440), 187–192 (Mar 2013). https://doi.org/10.1038/nature11971, http: //dx.doi.org/10.1038/nature11971

2013 doi
[17]

Nature Com- munications 15(1) (Jan 2024)

Ma, J., He, Y., Li, F., et al.: Segment anything in medical images. Nature Com- munications 15(1) (Jan 2024). https://doi.org/10.1038/s41467-024-44824-z , http://dx.doi.org/10.1038/s41467-024-44824-z

2024 doi
[18]

Frontiers in Neuroscience 15 (Aug 2021)

Meyer, M.I., de la Rosa, E., Pedrosa de Barros, N., Paolella, R., Van Leemput, K., Sima, D.M.: A contrast augmentation approach to improve multi-scanner gener- alization in mri. Frontiers in Neuroscience 15 (Aug 2021). https://doi.org/10. 3389/fnins.2021.708196, http://dx.doi....

2021
[19]

Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Figure 2), 6706–6716 (2020)

Misra, I., van der Maaten, L.: Self-supervised learning of pretext-invariant repre- sentations. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Figure 2), 6706–6716 (2020). https://doi.org/ 10.1109/CVPR42600.2020.00674

2020
[20]

Pathak, D., Krahenbuhl, P., Donahue, J., et al.: Context encoders: Feature learning by inpainting (2016), https://arxiv.org/abs/1604.07379

2016 arXiv
[21]

NeuroIm- age 50(3), 910–919 (Apr 2010)

Robinson, E.C., Hammers, A., Ericsson, A., et al.: Identifying population differ- ences in whole-brain structural networks: A machine learning approach. NeuroIm- age 50(3), 910–919 (Apr 2010). https://doi.org/10.1016/j.neuroimage.2010. 01.019, http://dx.doi.org/10.1016/j.neuro...

2010 doi
[22]

Roschewitz, M., Ribeiro, F.D.S., Xia, T., et al.: Robust image representations with counterfactual contrastive learning (2024), https://arxiv.org/abs/2409.10365

2024 arXiv
[23]

American Journal of Neuroradiology 38(6), 1103–1110 (Apr 2017)

Tanenbaum, L., Tsiouris, A., Johnson, A., et al.: Synthetic mri for clinical neu- roimaging: Results of the magnetic resonance image compilation (magic) prospec- tive, multicenter, multireader trial. American Journal of Neuroradiology 38(6), 1103–1110 (Apr 2017). https://doi.o...

2017 doi
[25]

Wang, Y., Li, Z., Mei, J., et al.: Swinmm: Masked multi-view with swin trans- formers for 3d medical image segmentation (2023),https://arxiv.org/abs/2307. 12591

2023
[26]

Nature Reviews Physics 3(8), 570–588 (Jun 2021)

Weiskopf, N., Edwards, L.J., Helms, G., et al.: Quantitative magnetic resonance imaging of brain anatomy and in vivo histology. Nature Reviews Physics 3(8), 570–588 (Jun 2021). https://doi.org/10.1038/s42254-021-00326-1 , http:// dx.doi.org/10.1038/s42254-021-00326-1

2021 doi
[27]

Wu, L., Zhuang, J., Chen, H.: Large-scale 3d medical image pre-training with geometric context priors (2024), https://arxiv.org/abs/2410.09890

2024 arXiv
[28]

Yang, Y., Gandhi, M., Wang, Y., et al.: A textbook remedy for domain shifts: Knowledge priors for medical image analysis (2024), https://arxiv.org/abs/ 2405.14839

2024 arXiv
[29]

Zbontar, J., Jing, L., Misra, I., et al.: Barlow twins: Self-supervised learning via redundancy reduction (2021), https://arxiv.org/abs/2103.03230

2021 arXiv

Pith tools

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

[1] [30]

Medical Image Analysis 67, 101840 (Jan 2021)

Zhou, Z., Sodha, V., Pang, J., Gotway, M.B., Liang, J.: Models genesis. Medical Image Analysis 67, 101840 (Jan 2021). https://doi.org/10.1016/j.media.2020. 101840, http://dx.doi.org/10.1016/j.media.2020.101840 12 Chalcroft et al. A Physics-based Signal Equations For each voxel we assume proton density (PD), longitudinal relaxation rate R1, transverse rela...

arXiv 2021

[2] [24]

Tang, Y., Yang, D., Li, W., et al.: Self-supervised pre-training of swin transformers for 3d medical image analysis (2022), https://arxiv.org/abs/2111.14791 SeqInv: Sequence-Invariant Contrastive Learning for 3D MRI 11

arXiv 2022

[3] [1]

https://doi.org/10.1101/2024.12.05.627056, http://dx.doi

Adams, R., Zhao, W., Hu, S., et al.: Ultimatesynth: Mri physics for pan-contrast ai (Dec 2024). https://doi.org/10.1101/2024.12.05.627056, http://dx.doi. org/10.1101/2024.12.05.627056

work page doi:10.1101/2024.12.05.627056 2024

[4] [2]

Asiedu, E.B., Kornblith, S., Chen, T., et al.: Decoder denoising pretraining for semantic segmentation (2022), https://arxiv.org/abs/2205.11423

work page Pith review arXiv 2022

[5] [3]

IEEE Transactions on Medi- cal Imaging 38(8), 1788–1800 (Aug 2019)

Balakrishnan, G., Zhao, A., Sabuncu, M.R., et al.: Voxelmorph: A learning frame- work for deformable medical image registration. IEEE Transactions on Medi- cal Imaging 38(8), 1788–1800 (Aug 2019). https://doi.org/10.1109/tmi.2019. 2897538, http://dx.doi.org/10.1109/TMI.2019.2897538

arXiv 2019

[6] [4]

Medical Image Analysis 86, 102789 (May 2023)

Billot, B., Greve, D.N., Puonti, O., et al.: SynthSeg: Segmentation of brain MRI scans of any contrast and resolution without retraining. Medical Image Analysis 86, 102789 (May 2023). https://doi.org/10.1016/j.media.2023.102789

arXiv 2023

[7] [5]

Borges, P., Shaw, R., Varsavsky, T., et al.: Acquisition-invariant brain mri segmen- tation with informative uncertainties (2021)

work page 2021

[8] [6]

Medical Image Analysis 92, 103058 (Feb 2024)

Borges, P., Shaw, R., Varsavsky, T., et al.: Acquisition-invariant brain mri seg- mentation with informative uncertainties. Medical Image Analysis 92, 103058 (Feb 2024). https://doi.org/10.1016/j.media.2023.103058, http://dx.doi. org/10.1016/j.media.2023.103058

arXiv 2024