Pith. sign in

REVIEW 4 major objections 6 minor 3 cited by

MRI-CORE: A Foundation Model for Magnetic Resonance Imaging

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read MRI-CORE, a self-supervised vision model trained on 6.9 million MRI slices, claims to outperform general-purpose segmentation foundation models such as SAM and MedSAM across 13 few-shot segmentation tasks, linear-probe classification, and…

desk verdict A well-ablated MRI foundation model with real few-shot gains; the short-pretraining claim is conditionally supported, and the zero-shot/FRD analyses need tightening. read the letter →

arxiv 2506.12186 v2 pith:TCV6A32V submitted 2025-06-13 eess.IV cs.AIcs.CVcs.LG

Haoyu Dong , Yuwen Chen , Hanxue Gu , Nicholas Konz , Yaqian Chen , Qihang Li , Maciej A. Mazurowski This is my paper · ORCID
classification eess.IVcs.AIcs.CVcs.LG
keywords magneticresonanceimagingfoundationmodelself-supervisedlearningDINOv2few-shotsegmentationzero-shottransferFréchetRadiomicDistance
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

MRI-CORE is a vision foundation model trained exclusively on MRI slices: more than 6 million slices from over 110,000 volumes covering 18 body locations and nine sequence types. The paper's central claim is that a slice-based, MRI-specific model, initialized with the weights of a general image segmentation model and then self-supervised trained for only a few epochs, produces representations that transfer to data-restricted medical tasks better than general-purpose segmentation foundation models. Across 13 few-shot segmentation tasks the model reports an average 3D Dice of 63.4, versus 59.1 for SAM and 48.1 for MedSAM, together with better frozen-feature classification and more semantically coherent zero-shot clusters. The paper also reports that the size of the gain over its own starting point grows as the downstream images become more similar to the pretraining distribution, measured by a radiomic distance metric. The authors note that the pretraining data comes from a single institution, excludes brain MRI, and uses a base-size vision transformer, so the claims are bounded to that regime.

What carries the argument

The central object is the DINOv2 student-teacher self-distillation objective, in which a student vision transformer is trained to match the embeddings of a teacher vision transformer under different image views, with masked-token reconstruction as an additional loss. MRI-CORE adapts this framework by initializing both student and teacher with the ViT-B image encoder of the Segment Anything Model, using slice-wise normalization, sampling without replacement per epoch, and a low learning rate so that only about four epochs of pretraining are run. The short schedule is load-bearing: the paper's ablations show performance peaks at epoch 4 and declines with longer training, which it interprets as preserving SAM's general visual knowledge while adapting it to MRI. The adapter-based few-shot fine-tuning protocol, with adapters inserted into the first and last two transformer blocks, is the mechanism through which transfer is measured.

What would settle it

Train MRI-CORE to checkpoints at epochs 2, 4, 8, 16, 32, and 64, then evaluate every downstream task on each checkpoint. If several tasks peak at an epoch other than 4, or if including brain MRI from a second institution shifts the optimum, the claim that a single short schedule is best would be undercut.

Watch

Extended reading notes

Core claim

The central claim is that a 2D vision transformer pretrained on a large, body-wide set of MRI slices with the DINOv2 self-supervised objective, starting from a Segment-Anything-initialized encoder and trained for a deliberately short schedule of about four epochs, becomes a strong MRI foundation model. The authors show this by fine-tuning only lightweight adapters in a five-slice few-shot setting on ten datasets covering bone, muscle, breast, organs, and tumor segmentation, and reporting that MRI-CORE outperforms SAM, MedSAM, and nnU-Net on average across 2D DSC, 3D DSC, and 3D NSD. They further show that the model's frozen features support linear classifiers for location, sequence, institution, disease, and severity tasks better than the same features from SAM or MedSAM, and that k-means clusters of its spatial features align more closely with anatomical structures in zero-shot settings. A distinctive secondary claim is that the transfer gain over the un-pretrained starting point tracks a distribution-similarity measure: the larger the Fréchet Radiomic Distance between pretraining and downstream images, the smaller the fine-tuning gain.

Load-bearing premise

The epoch-4 checkpoint was chosen using validation on two in-house datasets (bone and breast), and the paper assumes that same training length is near-optimal for all 13 downstream tasks; if the ideal pretraining length varies by anatomy or task, the reported average gains could reflect that selection.

Editorial extensions

If this is right

  • An MRI-only foundation model can be a stronger starting point for label-efficient MRI segmentation than natural-image or mixed-modality foundation models, so downstream users may need far fewer annotated slices for new tasks.
  • The success of a four-epoch schedule suggests that when initialization is strong, very short self-supervised adaptation can outperform long pretraining by retaining the inductive biases of the initial weights.
  • The Fréchet Radiomic Distance correlation gives a practical rule of thumb: tasks whose images lie close to the pretraining distribution are the ones most likely to benefit from MRI-CORE, so this distance could be used to predict where transfer will help.
  • Frozen MRI-CORE features carry usable signals for clinically relevant classification tasks such as anatomy, sequence, institution, and disease severity, which implies they can support linear-probe models trained with modest labels.

Reading between the lines

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

  • If the Fréchet Radiomic Distance relationship holds generally, it suggests a selection tool: before committing to a foundation model, compute the radiomic distance between its training data and the target dataset and choose the model with the smallest distance.
  • The exclusion of brain MRI from pretraining means the model's advantage on brain tasks is untested; a natural extension would be to measure whether adding a modest number of brain volumes closes any gap that remains on brain segmentation.
  • The same short-pretraining recipe could be tested on other imaging modalities such as CT, ultrasound, or pathology by initializing from a strong natural-image encoder and stopping after a handful of epochs; the paper's ablations predict this will beat longer training, but that extrapolation is not demonstrated here.
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.

Referee Report

4 major / 6 minor

Summary. The manuscript introduces MRI-CORE, a 2D vision foundation model for MRI trained with DINOv2-style self-supervised learning on Duke-110K (6.9 million slices from 116,806 volumes, single institution, no brain MRI). The model is initialized from SAM's ViT-B image encoder and trained for a deliberately short schedule (4 epochs). The authors evaluate few-shot segmentation with five annotated slices on 13 tasks across 10 datasets against SAM, MedSAM, nnU-Net, and a no-pretraining baseline; linear-probe classification on six tasks; and zero-shot segmentation via k-means clustering of patch embeddings. They also present an analysis relating the few-shot gain over SAM to the Fréchet Radiomic Distance between pre-training and downstream data. The central claims are that MRI-CORE outperforms the baselines on average by substantial margins, that short training from SAM initialization is preferable to longer training, and that transfer gains are predicted by FRD.

Significance. If the results hold, MRI-CORE is a valuable community resource: the model weights and code are released under a permissive license, the pre-training dataset is large and spans many body locations and sequence types, and the few-shot segmentation protocol is detailed and evaluated on multiple external public datasets. The paper also provides a useful ablation of training decisions (epochs, learning rate, normalization, initialization, and DINOv2 vs MAE) and a falsifiable, if exploratory, transfer-similarity hypothesis. The main quantitative claims are benchmarked against external models rather than only against the authors' own methods, which is a clear strength. However, the force of several claims is tempered by checkpoint selection on two in-house datasets, an oracle-style zero-shot evaluation, and a correlational FRD analysis with a small number of aggregated data points.

major comments (4)
  1. [Sec. 2.1, Sec. 4.2.2] The choice of the epoch 4 checkpoint for all downstream evaluations is made on Private-Bone and Private-Breast only, and these two datasets are also part of the 13-task segmentation benchmark. While the authors state that extended training 'reduced generalizability,' this conclusion is not tested on abdominal, cardiac, or spine tasks; if a later checkpoint is substantially better there, the released model would not be the best representative of the proposed short-pretraining strategy. Recomputing the average over the 11 tasks not used for checkpoint selection still leaves MRI-CORE ahead of SAM (about 3.7 vs 4.4 percentage points in 3D DSC), so the existence of a gain appears to survive, but the general claim about optimal training duration needs either an evaluation of epochs 2, 8, 16, 32, and 64 on the external tasks or a clearly conditional wording.
  2. [Sec. 2.4, Sec. 4.3.3, Fig. 5b] The quantitative zero-shot segmentation evaluation selects, for each image, the k-means cluster with the highest overlap with the ground-truth object ('find the cluster that overlaps with the target the most'), and the number of clusters k is chosen per dataset as the value giving the best DSC (32 for Private-Bone, 64 for AMOS-Liver). This is an oracle upper bound, not a zero-shot segmentation result: without labels one cannot know which cluster corresponds to the target, and per-dataset selection of k uses test information. The qualitative demonstrations remain interesting, but the quantitative claim of 'leading performance' in zero-shot segmentation is not established by the current protocol.
  3. [Sec. 2.2.2, Fig. 4] The FRD transfer analysis is based on 10 data points after aggregation, two of which (AMOS and HVSMR) combine multiple tasks and are therefore not independent; the p-values are uncorrected for multiple testing, and no confidence intervals or cross-validation are given. The sentence 'This directly establishes that the more similar a FSL fine-tuning set or test set is to the model's pre-training set, the more that the segmentation performance on the test set will improve' overstates what a correlational, post-hoc analysis can show. Please reframe the conclusion as an exploratory association, report the effective independent sample size or a bootstrap interval, and ideally validate the FRD-based prediction on held-out tasks.
  4. [Sec. 2.2.1, Tables A1-A13] The paper reports mean and standard deviation for each method and dataset but does not state the number of runs, the source of variance, or whether the same few-shot training slices and seeds were used across methods. With only five training slices per task, margins such as the 1.2-point 3D DSC difference over nnU-Net on Private-Breast or the -1.3-point deficit on SPIDER may be within run-to-run variability. Please add the number of repeated runs, seed sensitivity, and paired significance tests across datasets/tasks (or across repeated draws) to support the claim of a 'significant margin' over the baselines.
minor comments (6)
  1. [Sec. 1 (author affiliations)] The third affiliation is numbered 3 twice; the Department of Computer Science entry should be numbered 4.
  2. [Sec. 2.1] The phrase 'samling with replacement' is a typo for 'sampling with replacement.'
  3. [Sec. 2.2.1] The sentence 'Qualitatively, as shown in Fig. 2c' should refer to Fig. 2d, since Fig. 2c shows epoch-wise performance and Fig. 2d shows qualitative segmentation examples.
  4. [Fig. 4 caption] The phrase 'few-show learning' should be 'few-shot learning' in both panels of the caption.
  5. [Sec. 4.4 (LiverHccSeg)] The word 'segmetnation' is a typo for 'segmentation.'
  6. [Sec. 4.3.1] It is not stated explicitly whether SAM and MedSAM are fine-tuned with the same adapter protocol as MRI-CORE; please clarify this so that the reader can verify that the comparison is apples-to-apples.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; central performance claims are externally benchmarked and self-citations are not load-bearing.

full rationale

The core few-shot segmentation results are obtained by fine-tuning MRI-CORE and comparing against external baselines (SAM, MedSAM, nnU-Net, and a no-pretraining baseline) on 13 tasks. The pretraining strategy is an empirical modification of SAM initialization, so the reported gains are measured outcomes rather than consequences of a definition or a self-citation. The epoch-4 checkpoint was selected using separate development test sets for Private-Bone and Private-Breast and then applied to all downstream tasks; this is a model-selection step, not a fitted parameter that is renamed as a prediction. The FRD-based similarity analysis uses a metric proposed in prior work by the same group, but the correlation reported is a descriptive post-hoc analysis, not a prediction generated from the metric, and the metric itself is defined externally through radiomic features. No uniqueness theorem or ansatz is imported from the authors' prior work to force the model choice or the reported performance. The self-citations (Mazurowski et al. 2023, Gu et al. 2024, Gu et al. 2025, Konz et al. 2025) provide supporting empirical or methodological context and do not constitute a load-bearing self-referential chain. No specific equation or procedure in the paper reduces by construction to its own inputs.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical entities. The model itself is an artifact, not an invented entity. The main free parameters are pretraining hyperparameters and the zero-shot cluster count, all selected on development or per-dataset.

free parameters (5)
  • pretraining epochs = 4
    Chosen by epoch sweep (2, 4, 8, 16, 32, 64) on Private-Bone and Private-Breast development sets (Sec. 2.1).
  • pretraining learning rate = 4e-4
    Ablated against 4e-3 on the same development sets (Sec. 2.1).
  • SAM weight initialization = SAM ViT-B
    Ablated against random init and MAE; best on Private-Bone and Private-Breast (Sec. 2.1).
  • normalization strategy = slice-wise
    Ablated against volume-wise normalization; slice-wise performed better (Sec. 2.1).
  • zero-shot clustering k = 4 to 128 per dataset
    k is a hyperparameter; best per dataset reported (e.g., 32 for Private-Bone, 64 for AMOS-Liver), rather than a fixed value (Sec. 4.3.3, Fig. 5b).
assumptions (4)
  • standard math DINOv2 self-supervised learning produces transferable features on medical images
    Background method; assumed effective without re-derivation in this paper.
  • domain assumption Pretraining on 2D slices transfers to volumetric (3D) segmentation performance
    The model is 2D; 3D DSC is computed from slice predictions. The paper does not train a 3D model.
  • domain assumption Frechet Radiomic Distance (FRD) is a valid measure of distribution similarity for MRI
    Adopted from Konz et al. 2025 (same group) to support the transfer-similarity correlation.
  • domain assumption A 5-slice fine-tuning protocol with Adapter layers approximates label-efficient real-world use
    Evaluation protocol follows Gu et al. 2024; assumes the setting is practically meaningful.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MRI-CORE: A Foundation Model for Magnetic Resonance Imaging." pith.science (2026). https://pith.science/paper/TCV6A32V

@misc{pith2026250612186,
  author       = {Pith},
  title        = {Pith review of: MRI-CORE: A Foundation Model for Magnetic Resonance Imaging},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TCV6A32V}},
  note         = {Machine review of arXiv:2506.12186}
}
read the original abstract

The widespread use of Magnetic Resonance Imaging (MRI) in combination with deep learning shows promise for many high-impact automated diagnostic and prognostic tools. However, training new models requires large amounts of labeled data, a challenge due to high cost of precise annotations and data privacy. To address this issue, we introduce the MRI-CORE, a vision foundation model trained using more than 6 million slices from over 110 thousand MRI volumes across 18 body locations. Our experiments show notable improvements in performance over state-of-the-art methods in 13 data-restricted segmentation tasks, as well as in image classification, and zero-shot segmentation, showing the strong potential of MRI-CORE to enable data-efficient development of artificial intelligence models. We also present data on which strategies yield most useful foundation models and a novel analysis relating similarity between pre-training and downstream task data with transfer learning performance. Our model is publicly available with a permissive license.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. How Much MRI Preprocessing Is Enough? A Cost-Utility Study for Brain MRI Foundation Models

    cs.CV 2026-06 conditional novelty 6.0 of 10

    For MAE and JEPA brain MRI pretraining, moving from the cheapest stable preprocessing level (P2) to the best heavier pipeline buys only 3.4 and 1.8 percentage points of aggregate downstream utility, and most gains are...

  2. Are Vision Foundation Models Ready for Out-of-the-Box Medical Image Registration?

    eess.IV 2025-07 conditional novelty 6.0 of 10

    On a benchmark of four breast MRI registration tasks, the SAM encoder outperforms classical optimization methods for gross breast alignment but not for fine fibroglandular tissue, and medical pre-training did not help.

  3. Revisiting 2D Foundation Models for Scalable 3D Medical Image Classification

    cs.CV 2025-12 conditional novelty 5.0 of 10

    A frozen 2D vision foundation model with lightweight LoRA adapters and attention-based slice fusion achieves state-of-the-art 3D medical image classification across 12 tasks with about 1M trainable parameters per task.

Reference graph

Works this paper leans on

50 extracted references · 32 canonical work pages · cited by 3 Pith papers

  1. [1]

    write newline

    " write newline " cite write " FUNCTION editor.postfix editor num.names #1 > "( )" "( )" if FUNCTION editor.trans.postfix editor num.names #1 > "( )" "( )" if FUNCTION trans.postfix translator num.names #1 > "( )" "( )" if FUNCTION authors.editors.reflist.apa5 'field := 'dot := field num.names 'numnames := numnames 'format.num.names := format.num.names na...

  2. [2]

    sn-aps.bst

    FUNCTION identify.aps.version "sn-aps.bst" " [2024/07/19 v1.1 APS bibliography style]" * top ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key keywords month note number organization pages publisher school series title type url volume year eprint archive archivePrefix primaryClass adsurl adsnote version lab...

    2024

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap duplicate "7" = swap duplicate "8" = swap "9" = or or or or or or or or or FUNCTION n.separate 't := "" #0 'numnames := t empty not t #-1 #1 subs...

  4. [4]

    IEEE Transactions on Pattern Analysis and Machine Intelligence

    Azad R, Aghdam EK, Rauland A, et al (2024) Medical image segmentation review: The success of u-net. IEEE Transactions on Pattern Analysis and Machine Intelligence

    work page 2024

  5. [5]

    In: International Conference on Learning Representations

    Bi \'n kowski M, Sutherland DJ, Arbel M, et al (2018) Demystifying mmd gans. In: International Conference on Learning Representations

    work page 2018

  6. [6]

    doi:10.7937/K9/TCIA.2015.zF0vlOPv, ://doi.org/10.7937/K9/TCIA.2015.zF0vlOPv

    Bloch N, Madabhushi A, Huisman H, et al (2015) NCI-ISBI 2013 Challenge: Automated Segmentation of Prostate Structures . doi:10.7937/K9/TCIA.2015.zF0vlOPv, ://doi.org/10.7937/K9/TCIA.2015.zF0vlOPv

    doi:10.7937/k9/tcia.2015.zf0vlopv 2015

  7. [7]

    In: Proceedings of the IEEE/CVF international conference on computer vision, pp 9650--9660

    Caron M, Touvron H, Misra I, et al (2021) Emerging properties in self-supervised vision transformers. In: Proceedings of the IEEE/CVF international conference on computer vision, pp 9650--9660

    2021

  8. [8]

    Nature Medicine 30(3):850--862

    Chen RJ, Ding T, Lu MY, et al (2024) Towards a general-purpose foundation model for computational pathology. Nature Medicine 30(3):850--862

    work page 2024

Show all 50 references
  1. [9]

    doi:10.7937/K9/TCIA.2016.fADs26kG, ://doi.org/10.7937/K9/TCIA.2016.fADs26kG

    Fedorov A, Tempany C, Mulkern R, et al (2016) Data From QIN PROSTATE . doi:10.7937/K9/TCIA.2016.fADs26kG, ://doi.org/10.7937/K9/TCIA.2016.fADs26kG

    2016 doi

  2. [10]

    Scientific Data 11(1):264

    van der Graaf JW, van Hooff ML, Buckens CF, et al (2024) Lumbar spine segmentation in mr images: a dataset and a public benchmark. Scientific Data 11(1):264

    2024

  3. [11]

    Advances in neural information processing systems 33:21271--21284

    Grill JB, Strub F, Altch \'e F, et al (2020) Bootstrap your own latent-a new approach to self-supervised learning. Advances in neural information processing systems 33:21271--21284

    2020

  4. [12]

    Data in Brief 51:109662

    Gross M, Arora S, Huber S, et al (2023) Liverhccseg: A publicly available multiphasic mri dataset with liver and hcc tumor segmentations and inter-rater agreement analysis. Data in Brief 51:109662

    2023

  5. [13]

    arXiv preprint arXiv:240409957

    Gu H, Dong H, Yang J, et al (2024) How to build the best medical image segmentation algorithm using foundation models: a comprehensive empirical study with segment anything model. arXiv preprint arXiv:240409957

    2024

  6. [14]

    Medical Image Analysis p 103469

    Gu H, Colglazier R, Dong H, et al (2025) Segmentanybone: A universal model that segments any bone at any location on mri. Medical Image Analysis p 103469

    2025

  7. [15]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 16000--16009

    He K, Chen X, Xie S, et al (2022) Masked autoencoders are scalable vision learners. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 16000--16009

    2022

  8. [16]

    Advances in neural information processing systems 30

    Heusel M, Ramsauer H, Unterthiner T, et al (2017) Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems 30

    2017

  9. [17]

    In: International Conference on Machine Learning, PMLR, pp 2790--2799

    Houlsby N, Giurgiu A, Jastrzebski S, et al (2019) Parameter-efficient transfer learning for nlp. In: International Conference on Machine Learning, PMLR, pp 2790--2799

    2019

  10. [18]

    arXiv preprint arXiv:241016296

    Jha D, Susladkar OK, Gorade V, et al (2024) Cirrmri600+: Large scale mri collection and segmentation of cirrhotic liver. arXiv preprint arXiv:241016296

    2024

  11. [19]

    Advances in neural information processing systems 35:36722--36732

    Ji Y, Bai H, Ge C, et al (2022) Amos: A large-scale abdominal multi-organ benchmark for versatile medical image segmentation. Advances in neural information processing systems 35:36722--36732

    2022

  12. [20]

    In: Proceedings of the IEEE/CVF international conference on computer vision, pp 4015--4026

    Kirillov A, Mintun E, Ravi N, et al (2023) Segment anything. In: Proceedings of the IEEE/CVF international conference on computer vision, pp 4015--4026

    2023

  13. [21]

    https://universe.roboflow.com/kneemri-rqdae/kneemri-apmcr

    kneeMRI (2024) KNEEMRI Dataset . https://universe.roboflow.com/kneemri-rqdae/kneemri-apmcr

    2024

  14. [22]

    In: Medical Imaging with Deep Learning, ://openreview.net/forum?id=acNu3glOTh

    Konz N, Chen Y, Gu H, et al (2024) Rethinking perceptual metrics for medical image translation. In: Medical Imaging with Deep Learning, ://openreview.net/forum?id=acNu3glOTh

    2024

  15. [23]

    ://arxiv.org/abs/2412.01496, https://arxiv.org/abs/2412.01496 arXiv:2412.01496

    Konz N, Osuala R, Verma P, et al (2025) Fr\'echet radiomic distance (frd): A versatile metric for comparing medical imaging datasets. ://arxiv.org/abs/2412.01496, https://arxiv.org/abs/2412.01496 arXiv:2412.01496

    2025

  16. [24]

    In: The Twelfth International Conference on Learning Representations

    Li W, Yuille A, Zhou Z (2024) How well do supervised models transfer to 3d image segmentation. In: The Twelfth International Conference on Learning Representations

    2024

  17. [25]

    arXiv preprint arXiv:210610893

    Li Y, Wu M, Zhang Y, et al (2021) Piano: A parametric hand bone model from magnetic resonance imaging. arXiv preprint arXiv:210610893

    2021

  18. [26]

    Nature Medicine 30(3):863--874

    Lu MY, Chen B, Williamson DF, et al (2024) A visual-language foundation model for computational pathology. Nature Medicine 30(3):863--874

    2024

  19. [27]

    Nature Communications 15(1):654

    Ma J, He Y, Li F, et al (2024) Segment anything in medical images. Nature Communications 15(1):654

    2024

  20. [28]

    Medical Image Analysis 89:102918

    Mazurowski MA, Dong H, Gu H, et al (2023) Segment anything model for medical image analysis: an experimental study. Medical Image Analysis 89:102918

    2023

  21. [29]

    arXiv preprint arXiv:230407193

    Oquab M, Darcet T, Moutakanni T, et al (2023) Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:230407193

    2023

  22. [30]

    Medical Image Analysis 84:102704

    Osuala R, Kushibar K, Garrucho L, et al (2023) Data synthesis and adversarial networks: A review and meta-analysis in cancer imaging. Medical Image Analysis 84:102704

    2023

  23. [31]

    Jama 319(10):1024--1039

    Papanicolas I, Woskie LR, Jha AK (2018) Health care spending in the united states and other high-income countries. Jama 319(10):1024--1039

    2018

  24. [32]

    Nature Biomedical Engineering 9(4):521--538

    Sun Y, Wang L, Li G, et al (2025) A foundation model for enhancing magnetic resonance images and downstream segmentation, registration and diagnostic tasks. Nature Biomedical Engineering 9(4):521--538

    2025

  25. [33]

    arXiv preprint arXiv:230907778

    Vorontsov E, Bozkurt A, Casson A, et al (2023) Virchow: A million-slide digital pathology foundation model. arXiv preprint arXiv:230907778

    2023

  26. [34]

    arXiv preprint arXiv:250214064

    Wang S, Safari M, Li Q, et al (2025) Triad: Vision foundation model for 3d magnetic resonance imaging. arXiv preprint arXiv:250214064

    2025

  27. [35]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Woodland M, Castelo A, Taie MA, et al (2024) Feature extraction for generative medical imaging evaluation: New evidence against an evolving trend. In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    2024

  28. [36]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp 22873--22882

    Wu L, Zhuang J, Chen H (2024) Voco: A simple-yet-effective volume contrastive learning framework for 3d medical image analysis. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp 22873--22882

    2024

  29. [37]

    arXiv preprint arXiv:250217160

    Wu Y, Liu F, Yilmaz R, et al (2025) A pragmatic note on evaluating generative models with fr 'echet inception distance for retinal image synthesis. arXiv preprint arXiv:250217160

    2025

  30. [38]

    Bioengineering 11(10):1034

    Xu Y, Quan R, Xu W, et al (2024) Advances in medical image segmentation: a comprehensive review of traditional, deep learning and hybrid approaches. Bioengineering 11(10):1034

    2024

  31. [39]

    ://github.com/Hirriririir/Multimodal-Multiethnic-Thigh-Muscle-MRI-analysis

    Zhong H (2023) Hirriririir/multimodal-multiethnic-thigh-muscle-mri-analysis. ://github.com/Hirriririir/Multimodal-Multiethnic-Thigh-Muscle-MRI-analysis

    2023

  32. [40]

    arXiv preprint arXiv:211107832

    Zhou J, Wei C, Wang H, et al (2021) ibot: Image bert pre-training with online tokenizer. arXiv preprint arXiv:211107832

    2021

  33. [41]

    sn-basic.bst

    FUNCTION identify.basic.version "sn-basic.bst" " [2024/07/19 v1.1 bibliography style]" * top ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished institution journal key keywords month note number organization pages publisher school series ...

    2024

  34. [42]

    write newline

    " write newline "" before.all 'output.state := FUNCTION add.period duplicate empty 'skip "." * add.blank if FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap dupl...

  35. [43]

    write newline

    " write newline "" before.all 'output.state := FUNCTION output.doi doi empty skip "doi:" doi * "" * output if FUNCTION format.archive archivePrefix empty "" archivePrefix ":" * if FUNCTION format.primaryClass primaryClass empty "" " [" primaryClass * "] " * if FUNCTION format....

  36. [44]

    write newline

    " write newline "" before.all 'output.state := FUNCTION string.to.integer 't := t text.length 'k := #1 'char.num := t char.num #1 substring 's := s is.num s "." = or char.num k = not and char.num #1 + 'char.num := while char.num #1 - 'char.num := t #1 char.num substring FUNCTI...

  37. [45]

    write newline

    " write newline "" before.all 'output.state := FUNCTION string.to.integer 't := t text.length 'k := #1 'char.num := t char.num #1 substring 's := s is.num s "." = or char.num k = not and char.num #1 + 'char.num := while char.num #1 - 'char.num := t #1 char.num substring FUNCTI...

  38. [46]

    sn-nature.bst

    FUNCTION identify.nature.version "sn-nature.bst" " [2024/07/19 v1.1 bibliography style]" * top ENTRY address archive author booktitle chapter edition editor eprint howpublished institution journal key keywords month note number organization pages publisher school series title ...

    2024

  39. [47]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  40. [48]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  41. [49]

    sn-vancouver-num.bst

    FUNCTION identify.vancouver.version "sn-vancouver-num.bst" " [2024/07/19 v1.1 Vancouver bibliography style]" * top ENTRY address assignee author booktitle chapter cartographer day edition editor howpublished institution inventor journal key keywords month note number organizat...

    2024

  42. [50]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.