Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications

Dexuan Chen; Hao Chen; Juan Zhou; Juyoung Bae; Lingjie Yang; Moo Hyun Son; Neeraj Ramesh Mahboobani; Qiang Ye; Rui Zhang; Tao Li

REVIEW 4 major objections 6 minor 33 references

Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A multi-sequence MRI pretraining recipe that explicitly separates anatomy from scan contrast produces an encoder that ranks first on 39 of 44 downstream clinical benchmarks, including zero-shot transfer to unseen external datasets.

desk verdict PRISM is a genuinely large MRI pretraining resource with a broad evaluation, but the paper overclaims statistical significance and never verifies the anatomical fidelity of its synthetic translations. read the letter →

arxiv 2508.07165 v2 pith:HONPOWW4 submitted 2025-08-10 eess.IV cs.AIcs.CV

Zelin Qiu , Xi Wang , Zhuoyao Xie , Juan Zhou , Yu Wang , Lingjie Yang , Xinrui Jiang , Juyoung Bae

show 10 more authors

Moo Hyun Son Qiang Ye Dexuan Chen Rui Zhang Tao Li Neeraj Ramesh Mahboobani Varut Vardhanabhuti Xiaohui Duan Yinghua Zhao Hao Chen

This is my paper · ORCID

classification eess.IVcs.AIcs.CV

keywords MRIfoundationmodelmulti-sequencerepresentationlearninganatomy-invariantcontrastivedomaingeneralizationself-supervisedpretrainingmedicalimagesegmentationclinicalbenchmark

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

PRISM is a pretraining recipe for MRI that argues the main barrier to generalization is sequence heterogeneity: the same anatomy looks different across T1, T2, DWI, and other protocols, so models trained on one setting fail on others. The paper claims that splitting an encoder's features into anatomy-invariant and sequence-specific parts, and training that split with four self-supervised objectives on 336,476 volumes from ten body regions, solves the transfer problem. Across 44 downstream tasks—segmentation, diagnosis, grading, progression prediction, age estimation, registration, and report generation—PRISM ranks first on 39, with statistically significant margins, including zero-shot transfer to datasets never seen in pretraining or fine-tuning. If the claim holds, a single MRI encoder can be adapted to many clinical jobs with less labeled data and more robust performance across scanners, institutions, and acquisition protocols.

What carries the argument

The carrying mechanism is a dual-branch disentanglement module: features split into an anatomy-invariant subspace $f_{\mathrm{ana}}$ and a sequence-specific subspace $f_{\mathrm{seq}}$, trained by four concurrent pretext tasks. Masked image reconstruction enforces spatial coherence; P-space-guided image-to-image translation samples a latent protocol vector from a learned acquisition-parameter space, fuses it with $f_{\mathrm{seq}}$, and uses an adversarial discriminator to synthesize realistic alternate-contrast volumes; metadata prediction regresses TR, TE, and flip angle from $f_{\mathrm{seq}}$ and classifies body region from $f_{\mathrm{ana}}$; and the anatomy-invariant contrastive loss,

What would settle it

Run the P-space generator on a labeled segmentation dataset, compare Dice or surface distance between source and translated volumes to measure anatomical preservation, then ablate the anatomy-invariant contrastive loss. If synthetic translations alter anatomy, PRISM's external-transfer gains should shrink or vanish when translation fidelity is low, undercutting the disentanglement explanation.

Watch

Extended reading notes

Core claim

PRISM's core claim is that anatomy can be learned separately from contrast. The encoder partitions each volume into anatomical features $f_{\mathrm{ana}}$ shared across sequences and sequence-specific features $f_{\mathrm{seq}}$. A P-space-guided generator translates an input into synthetic volumes with different contrast while keeping the anatomical content, and the anatomical branch is pulled together by an anatomy-invariant contrastive loss while the sequence branch predicts acquisition metadata (TR, TE, flip angle) and body region; masked reconstruction keeps spatial structure intact. Trained on 336,476 volumes, the resulting encoder initializes task-specific heads and beats both from-sc

Load-bearing premise

The load-bearing premise, set up in the pretraining method, is that the P-space-guided translation changes contrast while preserving anatomy exactly, so the synthetic image's anatomical features are a valid positive pair for the source; the paper does not report a direct quantitative measure of translation anatomical fidelity.

Editorial extensions

If this is right

If PRISM's results hold, one pretrained encoder can replace task-specific MRI models for organ segmentation, lesion segmentation, disease grading, progression forecasting, age regression, cross-sequence registration, and report generation, all from a single initialization.
Zero-shot external transfer works: models fine-tuned on one dataset deploy to unseen datasets in the same anatomical region, improving Dice by roughly 1.1 to 2.5 points over the same architecture trained from scratch.
Data scale, not model capacity, is the main scaling lever: growing the pretraining corpus from 10k to 336k volumes improves downstream metrics, while increasing model size gives flat or negative returns.
Pretraining cuts fine-tuning cost: PRISM reaches higher Dice earlier and converges faster than from-scratch training, reducing the annotation and compute burden in clinical adaptation.
The four objectives are synergistic: masked reconstruction alone helps segmentation, metadata prediction helps sequence identification, and translation plus contrastive learning together produce the largest gains.

Reading between the lines

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

A balanced pretraining corpus would likely improve underrepresented organs: the paper's own discussion notes that over half of PRISM-336k volumes are knee scans and that gains concentrate on knee-related tasks; removing that imbalance is a natural next test.
The disentanglement account predicts a testable signature: translation fidelity should correlate with downstream robustness. Measuring anatomy preservation in the synthetic volumes, for example Dice between source and translated segmentation, would show whether the gains come from true anatomical invariance or simply from contrast augmentation.
Because the encoder transfers to language generation, jointly pretraining on MRI images paired with clinical reports—which the paper lists as future work—could push report quality further and may sharpen the anatomical representations through text supervision.
The external cohorts test zero-shot transfer across centers with the same anatomy and sequence types; the hardest remaining test is generalization to an entirely new MRI sequence or contrast not present in pretraining, where the learned P-space prior may or may not cover the protocol.

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

First, the key fact: PRISM is the largest multi-sequence MRI pretraining study I know of—336,476 volumes from 34 datasets, a 44-task benchmark, and a four-objective pretraining recipe. The scale and breadth are real, and the evaluation is unusually thorough: bootstrap CIs, external cohorts, ablations, and scaling studies. The paper also admits its own soft spots, such as the knee-dominated corpus and diminishing returns with model size. If the headline numbers hold, this is a useful resource for the field.

The integrated contribution is new: no prior work combines this corpus size with that pretext-task mix and evaluates across so many tasks. That alone makes it worth a serious look.

Now the soft spots, in order of importance.

First, the significance claims are overstated. In Section 2.2.3, OAI progression improvement is reported with p=0.064, and the text immediately says PRISM "consistently outperformed all compared models significantly." That p is not below 0.05, and the abstract's "39 out of 44 ... with statistical significance improvements" is not actually supported. This is a concrete error that should be fixed.

Second, the mechanism the paper sells—anatomy-invariant contrastive learning—is not actually validated. Equation (1) treats the synthetic translated image's anatomical features as a positive pair with the source image's anatomy. That only works if the translation preserves anatomy exactly. There is no structural consistency loss and no measurement of translation fidelity anywhere. The gains attributed to disentanglement might just be the usual benefit of data augmentation. Worse, the denominator wording ("negative samples from other subjects or sequences within the batch") is ambiguous; if same-subject different-sequence features are negatives, that actively contradicts the invariance claim. The stress-test note on this is on target: the core mechanism is unsupported, even though the empirical results may be genuine.

Third, the two closest predecessors cited in the introduction—Triad and MRI-CORE—are not in the benchmark. That weakens the SOTA claim.

Fourth, no code, weights, or data are released. For a foundation-model paper, that is a real reproducibility gap.

None of these are fatal. The central empirical conclusion—that MRI-specific pretraining at scale transfers across organs, sequences, and tasks—is credible and consistent with prior SSL findings. The paper deserves peer review, but the authors should be pushed to fix the significance language, add a translation-fidelity check or soften the mechanistic claims, compare against the cited predecessors, and ideally release at least the model weights.

Recommendation: send it to review, with major revision expected.

Referee Report

4 major / 6 minor

Summary. PRISM is an MRI foundation model pretrained on 336,476 multi-sequence volumes from 8 public and 26 private datasets. The pretraining combines four self-supervised tasks: masked volume reconstruction, a GAN-based 'P-space' cross-sequence translation, MRI metadata prediction (TR/TE/flip-angle regression plus body-part classification), and an anatomy-invariant InfoNCE contrastive loss on a disentangled anatomical feature branch. The encoder is a Swin Transformer and is adapted to segmentation (SwinUNETR), classification/regression (MLP heads), registration (TransMorph), and report generation (R2GenGPT with LLaMA). Evaluation covers 44 downstream tasks across 37 datasets split into Held-out, Independent, and External cohorts. The authors report first-rank results in 39/44 tasks, with bootstrap CIs and significance tests, plus scaling and ablation studies.

Significance. The scale and breadth of the benchmark are major strengths: 44 tasks, 37 datasets, three evaluation cohorts, and consistent reporting of 95% CIs with bootstrap significance tests. The external-transfer results (e.g., zero-shot CHAOS/MSD-Cardiac/PanSegData) are useful evidence that the pretrained representation generalizes beyond its training distribution. The 336k corpus and detailed data curation are also valuable contributions. However, the claimed mechanism of anatomical disentanglement is not yet supported: the contrastive positive pairs rely on unvalidated synthetic translations, and the loss formula as written can repel same-subject cross-sequence features. These issues are fixable but load-bearing for the paper's novelty claims, so I cannot recommend acceptance without revision.

major comments (4)

[4.2, Eq. (1)] Eq. (1) uses f^{n,i}_ana as the anchor and f^{n,i}_{s,ana} as the positive, with f^{m,k}_ana in the denominator. The text states that the denominator includes negative samples from 'other subjects or sequences within the batch.' With f^{m,k} indexed by subject m and sequence k, the case m=n, k≠i is the same subject's anatomical feature from another sequence. Making that a negative directly contradicts the stated goal of cross-sequence anatomical invariance. Please specify the negative set explicitly (e.g., restrict to m≠n), and if the implementation actually used same-subject cross-sequence negatives, rerun the affected ablations in Table 5.
[4.2, P-space guided image translation] The positive pair in Eq. (1) is (f_ana, f_{s,ana}), where f_{s,ana} comes from a synthetic image produced by P-space-guided translation. The paper reports no quantitative measure that the translated image preserves source anatomy; the latent-space discriminator only checks that re-encoded synthetic features are indistinguishable from real features, which can hold while the generator alters anatomy. Please add a structural consistency loss or an explicit fidelity metric (e.g., segmentation DSC or landmark error between source and synthetic volumes) to validate that the positive pair is indeed anatomically aligned. Without this, the ablation gains attributed to Con in Table 5 cannot be interpreted as anatomical invariance learning.
[2, STable 4] Fourteen of the 44 benchmarks are 'Held-out' tasks whose training data were also used in pretraining. The abstract's '39 out of 44' and the general 'generalization' language intermix these Held-out tasks with Independent and External cohorts. Since Held-out tasks do not measure distribution shift, please report first-rank counts separately for Held-out, Independent, and External, and base the generalization claims primarily on the Independent and External results. The Independent/External results appear strong, but the current reporting overstates the count of transfer benchmarks.
[2, baselines] The paper compares with MAE 3D, MoCo V3, SwinUNETR, and BrainSegFounder, but it does not state whether these were pretrained on the same PRISM-336k corpus with the same Swin backbone or whether released checkpoints were used. This is important because, for example, BrainSegFounder is trained only on brain data and may be disadvantaged on non-brain tasks. Please specify the pretraining data and recipe for each baseline, or add a controlled comparison where all baselines are pretrained on the same data with the same backbone.

minor comments (6)

[Fig. 1 and Section 4.2] Fig. 1(c) lists five pretext tasks (masked reconstruction, sequence translation, acquisition parameter prediction, anatomical region classification, contrastive learning), while the abstract and Section 4.2 say four objectives. The metadata task has two subheads, so the count should be made consistent.
[Table 1 caption] Table 1 caption contains a typo: 'PI-AI' should be 'PI-CAI'.
[4.6 Statistical Analysis] The sentence 'the Wilcoxon test is used for ACC and DSC using bootstrapping' is unclear. Please clarify whether bootstrapping is used only for confidence intervals and Wilcoxon signed-rank for paired metric comparisons, or whether a bootstrap p-value is computed.
[4.2, P-space translation details] The generator, decoder, MLP that maps z to fp, and the latent discriminator are described only at a high level. To make the method reproducible, please report the GAN loss weighting, the architecture of the translation decoder, and the discriminator design.
[5 Data Availability] 'Source data are provided with this paper' conflicts with the statement that private data is 'currently unavailable to the public.' Please clarify what source data are provided and what can be shared.
[2.2.3 Progression Prediction] The OAI progression improvement has p=0.064 (STable 14, no star), but the text says PRISM 'consistently outperformed all compared models significantly.' Please qualify this claim.

Circularity Check

2 steps flagged · score 4.0 of 10

Core anatomy-invariance objective is self-referential (Eq. 1), but the headline empirical claim rests on external cohorts and direct measurements, so only partial circularity.

self definitional [Section 4.2, Eq. (1) (Anatomy invariant contrastive learning)]
"where τ is a temperature scaling factor, and the denominator includes negative samples from other subjects or sequences within the batch. This formulation encourages anatomical features to be invariant across varying contrast domains"

As written, the denominator treats 'other sequences within the batch' as negatives. If the batch contains two sequences of the same subject, the loss actively repels that subject's anatomical features across sequences—the opposite of the claimed cross-sequence invariance. The only attraction is between an original and its own synthetic translation, whose 'anatomical' status is asserted rather than verified. Thus the property 'anatomical invariance' is not derived from the loss; it is imposed by naming the positive pair anatomical and by a verbal interpretation of the denominator. The robustness claims in Sections 2-3 and the ablation in Table 5 attribute gains to this mechanism, so the loading is on an equation whose stated negative set can contradict the intended invariance.
self definitional [Section 4.2, P-space guided image translation]
"This vector fp is then fused with the sequence-specific feature fseq, and the combined representation is subsequently integrated with fana through a convolutional layer to preserve anatomical integrity. ... To further enforce realism and parameter fidelity, the synthetic image is re-encoded, and a latent-space discriminator evaluates whether the resulting feature distribution is indistinguishable from that of real images."

The positive pair in Eq. (1) is defined as f_ana(original) versus f_ana(synthetic). The synthetic image is produced by the model itself and is supervised only by a feature-distribution discriminator, with no structural-consistency or segmentation-fidelity loss. Consequently, 'anatomical integrity' is asserted rather than verified; the model can satisfy the contrastive objective by making original and synthetic features similar even if the translation altered anatomy. The learned subspace is therefore 'anatomical' by labeling/construction, not by any independent anatomical correspondence.

full rationale

The paper's headline empirical claim—first-rank results on 39/44 downstream benchmarks—is not a fitted parameter renamed as a prediction. The External cohort is excluded from both pretraining and fine-tuning, the Independent cohort excludes pretraining overlap, and baselines are evaluated under the same splits and protocols, so the central result has independent empirical content. The circularity that does exist is localized to the pretraining mechanism: the anatomy-invariant contrastive loss (Eq. 1) defines 'anatomical features' as those that are similar between an original image and a synthetic translation generated by the same network, with no anatomical fidelity check; and its denominator can include same-subject cross-sequence negatives, which contradicts the claimed invariance. These issues make the disentanglement claim self-referential, and because the ablations (Table 5) and discussion attribute robustness to this mechanism, they are load-bearing for the explanation, though not for the raw measurements. The supplementary self-citation for knee data splits ([31]) is a minor data-protocol reference and not load-bearing. Overall: partial circularity in the core mechanism, not in the empirical comparison.

Assumptions & free parameters 4 free parameters · 6 assumptions · 2 invented entities

The central claim is an empirical transfer measurement, so this ledger captures the hand-set values and assumed mechanisms the pretext objectives rely on: unreported loss weights and temperatures, the assumed extractability of acquisition metadata, and the assumed anatomical fidelity of translated images. The two invented constructs (P-space and the f_ana/f_seq split) carry the paper's mechanistic narrative but have no falsifiable handle outside the paper. No pretraining constant is fitted to the downstream results, which keeps circularity moderate rather than high.

free parameters (4)

InfoNCE temperature tau (Eq. 1) = not reported
Sets the sharpness of the anatomy-invariant contrastive loss; hand-chosen and undisclosed, yet it directly shapes the main novel objective.
Pretext loss weights (reconstruction, translation, metadata, contrastive) = not reported
The four objectives are combined end-to-end in Section 4.2, but the balancing weights are not given. The ablation (Table 5) shows the mix matters (up to 1.4 point swings), so this is a load-bearing hand-set choice.
Masking volume ratio s = 30%
Stated in Section 4.1, following prior work [58]; defines the difficulty of the reconstruction pretext and is chosen rather than derived.
P-space latent dimension and translation MLP/decoder design = not reported
Defines the capacity of the generator that produces synthetic contrast variants; without these values the core novelty cannot be re-implemented.

assumptions (6)

ad hoc to paper Encoder output features can be partitioned into orthogonal anatomical and sequence-specific subspaces whose sum still captures full MRI content.
Used in Section 4.2 to justify the dual-branch disentanglement and both generative and contrastive pretext losses; no identifiability or orthogonality argument is provided.
ad hoc to paper Translated synthetic images preserve source anatomy, so f_{s,ana} and f_{ana} form valid positive pairs in Eq. 1.
Load-bearing for the anatomy-invariant contrastive loss; no anatomical-fidelity metric for translated images is reported.
domain assumption TR, TE, and flip angle are accurately extractable for all 336,476 pretraining volumes and are valid proxies for image contrast.
Stated in Section 4.1; many public datasets are distributed as NIfTI without complete DICOM headers, so parameter availability is assumed rather than verified.
domain assumption Identical training hyperparameters across comparison models guarantee fair baselines.
Claimed in Section 4.3, but baseline training budgets, augmentations, and the pretraining data for MAE 3D and MoCo V3 are not reported.
domain assumption The Held-out/Independent/External split design prevents data leakage between pretraining and fine-tuning.
The authors say splits were fixed before training and adopt BrainSegFounder's BraTS fold list; for other public datasets, leakage cannot be independently verified because scan-level identifiers are not disclosed.
standard math Swin Transformer, SwinUNETR, InfoNCE, and adversarial training behave as in their source papers.
Background machinery; no re-derivation is needed, but correctness of these components is imported.

invented entities (2)

P-space (latent acquisition-parameter space)
purpose: A latent code sampled from N(0,I), mapped through an MLP, and fused with sequence features to drive contrast-controlled image translation during pretraining.
Internal construct with no external falsifiable handle: the paper does not require the sampled code to match any real acquisition's TR/TE/FA, so the translation pipeline's claims cannot be checked independently.
Dual-branch anatomical/sequence feature split (f_ana, f_seq)
purpose: Supports the claim that anatomical and acquisition-dependent information are linearly separable in the encoder output and that contrastive pull on f_ana yields robustness.
Postulated representational structure; its correctness is inferred only from downstream gains and the paper's own ablations, with no direct probing or controlled experiment isolating f_ana.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications." pith.science (2026). https://pith.science/paper/HONPOWW4

@misc{pith2026250807165,
  author       = {Pith},
  title        = {Pith review of: Large-scale Multi-sequence Pretraining for Generalizable MRI Analysis in Versatile Clinical Applications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HONPOWW4}},
  note         = {Machine review of arXiv:2508.07165}
}

read the original abstract

Multi-sequence Magnetic Resonance Imaging (MRI) offers remarkable versatility, enabling the distinct visualization of different tissue types. Nevertheless, the inherent heterogeneity among MRI sequences poses significant challenges to the generalization capability of deep learning models. These challenges undermine model performance when faced with varying acquisition parameters, thereby severely restricting their clinical utility. In this study, we present PRISM, a foundation model PRe-trained with large-scale multI-Sequence MRI. We collected a total of 64 datasets from both public and private sources, encompassing a wide range of whole-body anatomical structures, with scans spanning diverse MRI sequences. Among them, 336,476 volumetric MRI scans from 34 datasets (8 public and 26 private) were curated to construct the largest multi-organ multi-sequence MRI pretraining corpus to date. We propose a novel pretraining paradigm that disentangles anatomically invariant features from sequence-specific variations in MRI, while preserving high-level semantic representations. We established a benchmark comprising 44 downstream tasks, including disease diagnosis, image segmentation, registration, progression prediction, and report generation. These tasks were evaluated on 32 public datasets and 5 private cohorts. PRISM consistently outperformed both non-pretrained models and existing foundation models, achieving first-rank results in 39 out of 44 downstream benchmarks with statistical significance improvements. These results underscore its ability to learn robust and generalizable representations across unseen data acquired under diverse MRI protocols. PRISM provides a scalable framework for multi-sequence MRI analysis, thereby enhancing the translational potential of AI in radiology. It delivers consistent performance across diverse imaging protocols, reinforcing its clinical applicability.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 29 canonical work pages

[1]

Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Transactions on Medical Imaging , 37(11):2514–2525, 2018

Olivier Bernard, Alain Lalande, Clement Zotti, Frederick Cerven ansky, Xin Yang, Pheng-Ann Heng, Irem Cetin, Karim Lekadir, Oscar Camara, Miguel Angel Gonzalez Ballester, et al. Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Transactions on Medical Imaging , 37(11):2514–2525, 2018

work page 2018
[2]

The alzheimer’s d isease neuroimaging initiative

Susanne G Mueller, Michael W Weiner, Leon J Thal, Ronald C Petersen , Cliﬀord Jack, William Jagust, John Q Trojanowski, Arthur W Toga, and Laurel Beckett. The alzheimer’s d isease neuroimaging initiative. Neuroimaging Clinics , 15(4):869–877, 2005

work page 2005
[3]

Amos: A large-scale abdominal multi-organ benchmark for ver satile medical image segmentation

Yuanfeng Ji, Haotian Bai, Jie Yang, Chongjian Ge, Ye Zhu, Ruimao Zhang, Zh en Li, Lingyan Zhang, Wanling Ma, Xiang Wan, et al. Amos: A large-scale abdominal multi-organ benchmark for ver satile medical image segmentation. arXiv preprint arXiv:2206.08023 , 2022

arXiv 2022
[4]

A tumour and liver automatic seg- mentation (ATLAS) dataset on contrast-enhanced magnetic resonance imaging for hepatocellular carcinoma

F´ elix Quinton, Romain Popoﬀ, Beno ˆ ıt Presles, Sarah Leclerc, Fabrice Meriaudeau, Guillaume Nodari, Olivier Lopez, Julie Pellegrinelli, Olivier Chevallier, Dominique Ginh ac, et al. A tumour and liver automatic seg- mentation (ATLAS) dataset on contrast-enhanced magnetic resonance imaging for hepatocellular carcinoma. Data, 8(5):79, 2023

work page 2023
[5]

The RSNA-ASNR-MICCAI BraTS 2021 benchmark on brain tumor segmentation and radiogenomic classiﬁcation

Ujjwal Baid, Satyam Ghodasara, Suyash Mohan, Michel Bilello, Evan Calabre se, Errol Colak, Keyvan Fara- hani, Jayashree Kalpathy-Cramer, Felipe C Kitamura, Sarthak Pati, et al . The RSNA-ASNR-MICCAI BraTS 2021 benchmark on brain tumor segmentation and radiogenomic classiﬁcation . arXiv preprint arXiv:2107.02314, 2021

arXiv 2021
[6]

Minimizing estimated risks on unlabele d data: A new formulation for semi- supervised medical image segmentation

Fuping Wu and Xiahai Zhuang. Minimizing estimated risks on unlabele d data: A new formulation for semi- supervised medical image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence , 45(5):6021–6036, 2023

work page 2023
[7]

Emre Kavur, N

A. Emre Kavur, N. Sinem Gezer, Mustafa Barı¸ s, Sinem Aslan, Pierr e-Henri Conze, Vladimir Groza, Duc Duy Pham, Soumick Chatterjee, Philipp Ernst, Sava¸ s ¨Ozkan, Bora Baydar, Dmitry Lachinov, Shuo Han, Josef Pauli, Fabian Isensee, Matthias Perkonigg, Rachana Sathish, Ronnie Rajan , Debdoot Sheet, Gurbandurdy Dovletov, Oliver Speck, Andreas N¨ urnberger, ...

work page 2021
[8]

A machine learning approach to radiogenomics of breas t cancer: a study of 922 subjects and 529 DCE-MRI features

Ashirbani Saha, Michael R Harowicz, Lars J Grimm, Connie E Kim, Sujata V Ghate, Ruth Walsh, and Maciej A Mazurowski. A machine learning approach to radiogenomics of breas t cancer: a study of 922 subjects and 529 DCE-MRI features. British Journal of Cancer , 119(4):508–516, 2018

work page 2018

Show all 33 references

[9]

Duke liver dataset: A publicly available liver MRI dataset w ith liver segmentation masks and series labels

Jacob A Macdonald, Zhe Zhu, Brandon Konkel, Maciej A Mazurowski, Walte r F Wiggins, and Mustafa R Bashir. Duke liver dataset: A publicly available liver MRI dataset w ith liver segmentation masks and series labels. Radiology: Artiﬁcial Intelligence , 5(5):e220275, 2022. 16

2022
[10]

Emid ec: a database usable for the automatic evaluation of myocardial infarction from delayed-enhancement cardiac MRI

Alain Lalande, Zhihao Chen, Thomas Decourselle, Abdul Qayyum, Thibau t Pommier, Luc Lorgis, Ezequiel de La Rosa, Alexandre Cochet, Yves Cottin, Dominique Ginhac, et al. Emid ec: a database usable for the automatic evaluation of myocardial infarction from delayed-enhancement ca...

2020
[11]

fastMRI: A pu blicly available raw k-space and dicom dataset of knee images for accelerated mr image reconstruction using machine learning

Florian Knoll, Jure Zbontar, Anuroop Sriram, Matthew J Muckley, Mary Bruno, Aaron Defazio, Marc Parente, Krzysztof J Geras, Joe Katsnelson, Hersh Chandarana, et al. fastMRI: A pu blicly available raw k-space and dicom dataset of knee images for accelerated mr image reconstruct...

2020
[12]

HaN-Seg: The head and neck organ-at-risk ct and mr segmentation dataset

Gaˇ sper Podobnik, Primoˇ z Strojan, Primoˇ z Peterlin, BulatIbragimov, and Tomaˇ z Vrtovec. HaN-Seg: The head and neck organ-at-risk ct and mr segmentation dataset. Medical Physics, 50(3):1917–1927, 2023

1917
[13]

IXI dataset – brai n development

Biomedical Image Analysis Group, Imperial College London. IXI dataset – brai n development. https:// brain-development.org/ixi-dataset/, 2024. Accessed: 2025-04-15

2024
[14]

SDR-Former: A siamese dual-resolution transformer for liver lesion classiﬁcation using 3d mu lti-phase imaging

Meng Lou, Hanning Ying, Xiaoqing Liu, Hong-Yu Zhou, Yuqin Zhang, and Yizhou Yu. SDR-Former: A siamese dual-resolution transformer for liver lesion classiﬁcation using 3d mu lti-phase imaging. Neural Networks, page 107228, 2025

2025
[15]

A large-scale multicenter breast cancer DCE-MRI benchmark dataset with expert segmentations

Lidia Garrucho, Kaisar Kushibar, Claire-Anne Reidel, Smriti Joshi , Richard Osuala, Apostolia Tsirikoglou, Maciej Bobowicz, Javier del Riego, Alessandro Catanese, Katarzyna Gwo ´ zdziewicz, Maria-Laura Cosaka, Pasant M Abo-Elhoda, Sara W Tantawy, Shorouq S Sakrana, Norhan O Sh...

2025
[16]

Multivariate mixture model for myocardial segmentation combining multi-source images

Xiahai Zhuang. Multivariate mixture model for myocardial segmentation combining multi-source images. IEEE Transactions on Pattern Analysis and Machine Intelligence , 41(12):2933–2946, 2018

2018
[17]

Deep-learni ng-assisted diagnosis for knee magnetic resonance imaging: development and retrospective validation of MRNet

Nicholas Bien, Pranav Rajpurkar, Robyn L Ball, Jeremy Irvin, Allis on Park, Erik Jones, Michael Bereket, Bhavik N Patel, Kristen W Yeom, Katie Shpanskaya, et al. Deep-learni ng-assisted diagnosis for knee magnetic resonance imaging: development and retrospective validation of ...

2018
[18]

The medical segmentation decathlon

Michela Antonelli, Annika Reinke, Spyridon Bakas, Keyvan Farahani, Annette Kopp-Schneider, Bennett A Landman, Geert Litjens, Bjoern Menze, Olaf Ronneberger, Ronald M Summ ers, et al. The medical segmentation decathlon. Nature Communications, 13(1):4128, 2022

2022
[19]

The osteoarthriti s initiative

Michael Nevitt, David Felson, and Gayle Lester. The osteoarthriti s initiative. Protocol for the Cohort Study , 1:2, 2006

2006
[20]

Aut omated segmentation of knee bone and cartilage combining statistical shape knowledge and convolutional neural networks: Data from the osteoarthritis initiative

Felix Ambellan, Alexander Tack, Moritz Ehlke, and Stefan Zachow. Aut omated segmentation of knee bone and cartilage combining statistical shape knowledge and convolutional neural networks: Data from the osteoarthritis initiative. Medical Image Analysis , 52:109–118, 2019

2019
[21]

Oasis-3: longitudinal neu- roimaging, clinical, and cognitive dataset for normal aging and Alzheimer dise ase

Pamela J LaMontagne, Tammie LS Benzinger, John C Morris, Sarah Keefe, Rus s Hornbeck, Chengjie Xiong, Elizabeth Grant, Jason Hassenstab, Krista Moulder, Andrei G Vlassenko, e t al. Oasis-3: longitudinal neu- roimaging, clinical, and cognitive dataset for normal aging and Alzhe...

2019
[22]

OpenBhB: a large-scale multi-site brain MRI data-set for age pred iction and debiasing

Benoit Dufumier, Antoine Grigis, Julie Victor, Corentin Ambroise, Vincent Frouin, and Edouard Duches- nay. OpenBhB: a large-scale multi-site brain MRI data-set for age pred iction and debiasing. NeuroImage, 263:119637, 2022

2022
[23]

Large-scale mul ti-center CT and MRI segmentation of pancreas with deep learning

Zheyuan Zhang, Elif Keles, Gorkem Durak, Yavuz Taktak, Onkar Suslad kar, Vandan Gorade, Debesh Jha, Asli C Ormeci, Alpay Medetalibeyoglu, Lanhong Yao, et al. Large-scale mul ti-center CT and MRI segmentation of pancreas with deep learning. Medical Image Analysis , 99:103382, 2025

2025
[24]

Art iﬁcial intelligence and radiologists in prostate cancer detection on MRI (PI-CAI): an international, paired, non-inferiority, conﬁrmatory study

Anindo Saha, Joeran S Bosma, Jasper J Twilt, Bram van Ginneken, Anders Bjartell, Anwar R Padhani, David Bonekamp, Geert Villeirs, Georg Salomon, Gianluca Giannarini, et al. Art iﬁcial intelligence and radiologists in prostate cancer detection on MRI (PI-CAI): an international,...

2024
[25]

The Parkinson’s progression markers initiative (PPMI)–establishing a PD biomarker c ohort

Kenneth Marek, Sohini Chowdhury, Andrew Siderowf, Shirley Las ch, Christopher S Coﬀey, Chelsea Caspell- Garcia, Tanya Simuni, Danna Jennings, Caroline M Tanner, John Q Trojano wski, et al. The Parkinson’s progression markers initiative (PPMI)–establishing a PD biomarker c ohor...

2018
[26]

E valuation of prostate segmentation algorithms for MRI: the PROMISE12 challenge

Geert Litjens, Robert Toth, Wendy Van De Ven, Caroline Hoeks, Sjoe rd Kerkstra, Bram Van Ginneken, Graham Vincent, Gwenael Guillard, Neil Birbeck, Jindang Zhang, et al. E valuation of prostate segmentation algorithms for MRI: the PROMISE12 challenge. Medical Image Analysis , 1...

2014
[27]

Prostate158-an expert-annotated 3T mri dataset and algorithm for prostate cancer detection

Lisa C Adams, Marcus R Makowski, G¨ unther Engel, Maximilian Rattund e, Felix Busch, Patrick Asbach, Stefan M Niehues, Shankeeth Vinayahalingam, Bram van Ginneken, Geert Li tjens, et al. Prostate158-an expert-annotated 3T mri dataset and algorithm for prostate cancer detection...

2022
[28]

Skm-tea: A dataset for accelerated MRI reconstruction with dense image labels for quantitati ve clinical evaluation

Arjun D Desai, Andrew M Schmidt, Elka B Rubin, Christopher M Sand ino, Marianne S Black, Valentina Mazzoli, Kathryn J Stevens, Robert Boutin, Christopher R´ e, Garry E Gold, et al. Skm-tea: A dataset for accelerated MRI reconstruction with dense image labels for quantitati ve ...

2022 arXiv
[29]

Lumbar spine segmentation in MR images: a dataset and a public benchmark

Jasper W van der Graaf, Miranda L van Hooﬀ, Constantinus FM Buckens, Mat thieu Rutten, Job LC van Susante, Robert Jan Kroeze, Marinus de Kleuver, Bram van Ginneken, and Nikolas Lessmann. Lumbar spine segmentation in MR images: a dataset and a public benchmark. Scientiﬁc Data ,...

2024
[30]

TotalSegmenta- tor MRI: Sequence-independent segmentation of 59 anatomical structur es in MR images

Tugba Akinci D’Antonoli, Lucas K Berger, Ashraya K Indrakanti, Nathan Vish wanathan, Jakob Weiß, Matthias Jung, Zeynep Berkarda, Alexander Rau, Marco Reisert, Thomas K¨ ustner, et al. TotalSegmenta- tor MRI: Sequence-independent segmentation of 59 anatomical structur es in MR ...

2024 arXiv
[31]

Learning co-plane attention across MRI sequences for diagnosin g twelve types of knee abnormalities

Zelin Qiu, Zhuoyao Xie, Huangjing Lin, Yanwen Li, Qiang Ye, Menghong W ang, Shisi Li, Yinghua Zhao, and Hao Chen. Learning co-plane attention across MRI sequences for diagnosin g twelve types of knee abnormalities. Nature Communications, 15(1):7637, 2024

2024
[32]

Enhancing MRI -based classiﬁcation of Alzheimer’s disease with explainable 3D hybrid compact convolutional t ransformers

Arindam Majee, Avisek Gupta, Sourav Raha, and Swagatam Das. Enhancing MRI -based classiﬁcation of Alzheimer’s disease with explainable 3D hybrid compact convolutional t ransformers. In 2024 International Joint Conference on Neural Networks (IJCNN) , pages 1–8. IEEE, 2024

2024
[33]

Knee osteoarthritis classiﬁ cation using 3D CNN and MRI

Carmine Guida, Ming Zhang, and Juan Shan. Knee osteoarthritis classiﬁ cation using 3D CNN and MRI. Applied Sciences, 11(11):5196, 2021. 18

2021

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Transactions on Medical Imaging , 37(11):2514–2525, 2018

Olivier Bernard, Alain Lalande, Clement Zotti, Frederick Cerven ansky, Xin Yang, Pheng-Ann Heng, Irem Cetin, Karim Lekadir, Oscar Camara, Miguel Angel Gonzalez Ballester, et al. Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Transactions on Medical Imaging , 37(11):2514–2525, 2018

work page 2018

[2] [2]

The alzheimer’s d isease neuroimaging initiative

Susanne G Mueller, Michael W Weiner, Leon J Thal, Ronald C Petersen , Cliﬀord Jack, William Jagust, John Q Trojanowski, Arthur W Toga, and Laurel Beckett. The alzheimer’s d isease neuroimaging initiative. Neuroimaging Clinics , 15(4):869–877, 2005

work page 2005

[3] [3]

Amos: A large-scale abdominal multi-organ benchmark for ver satile medical image segmentation

Yuanfeng Ji, Haotian Bai, Jie Yang, Chongjian Ge, Ye Zhu, Ruimao Zhang, Zh en Li, Lingyan Zhang, Wanling Ma, Xiang Wan, et al. Amos: A large-scale abdominal multi-organ benchmark for ver satile medical image segmentation. arXiv preprint arXiv:2206.08023 , 2022

arXiv 2022

[4] [4]

A tumour and liver automatic seg- mentation (ATLAS) dataset on contrast-enhanced magnetic resonance imaging for hepatocellular carcinoma

F´ elix Quinton, Romain Popoﬀ, Beno ˆ ıt Presles, Sarah Leclerc, Fabrice Meriaudeau, Guillaume Nodari, Olivier Lopez, Julie Pellegrinelli, Olivier Chevallier, Dominique Ginh ac, et al. A tumour and liver automatic seg- mentation (ATLAS) dataset on contrast-enhanced magnetic resonance imaging for hepatocellular carcinoma. Data, 8(5):79, 2023

work page 2023

[5] [5]

The RSNA-ASNR-MICCAI BraTS 2021 benchmark on brain tumor segmentation and radiogenomic classiﬁcation

Ujjwal Baid, Satyam Ghodasara, Suyash Mohan, Michel Bilello, Evan Calabre se, Errol Colak, Keyvan Fara- hani, Jayashree Kalpathy-Cramer, Felipe C Kitamura, Sarthak Pati, et al . The RSNA-ASNR-MICCAI BraTS 2021 benchmark on brain tumor segmentation and radiogenomic classiﬁcation . arXiv preprint arXiv:2107.02314, 2021

arXiv 2021

[6] [6]

Minimizing estimated risks on unlabele d data: A new formulation for semi- supervised medical image segmentation

Fuping Wu and Xiahai Zhuang. Minimizing estimated risks on unlabele d data: A new formulation for semi- supervised medical image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence , 45(5):6021–6036, 2023

work page 2023

[7] [7]

Emre Kavur, N

A. Emre Kavur, N. Sinem Gezer, Mustafa Barı¸ s, Sinem Aslan, Pierr e-Henri Conze, Vladimir Groza, Duc Duy Pham, Soumick Chatterjee, Philipp Ernst, Sava¸ s ¨Ozkan, Bora Baydar, Dmitry Lachinov, Shuo Han, Josef Pauli, Fabian Isensee, Matthias Perkonigg, Rachana Sathish, Ronnie Rajan , Debdoot Sheet, Gurbandurdy Dovletov, Oliver Speck, Andreas N¨ urnberger, ...

work page 2021

[8] [8]

A machine learning approach to radiogenomics of breas t cancer: a study of 922 subjects and 529 DCE-MRI features

Ashirbani Saha, Michael R Harowicz, Lars J Grimm, Connie E Kim, Sujata V Ghate, Ruth Walsh, and Maciej A Mazurowski. A machine learning approach to radiogenomics of breas t cancer: a study of 922 subjects and 529 DCE-MRI features. British Journal of Cancer , 119(4):508–516, 2018

work page 2018