Pith. sign in

REVIEW 4 major objections 4 minor 41 references

PathPT converts weak slide-level labels into tile-level supervision for rare cancer, beating four MIL baselines in few-shot subtyping.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

PathPT improves few-shot rare cancer subtyping by using zero-shot vision-language models to create tile-level pseudo-labels and learning prompt tokens with spatial context, outperforming standard MIL baselines when the backbone has strong zero-shot grounding.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection PathPT is a useful few-shot recipe for VL-based WSI subtyping with a genuinely valuable pediatric benchmark, but the consistency claim is overblown and the few-shot protocol has an unresolved label-leakage ambiguity. the 4 major comments →

arxiv 2508.15904 v1 pith:CAESZVON submitted 2025-08-21 cs.CV

Boosting Pathology Foundation Models via Few-shot Prompt-tuning for Rare Cancer Subtyping

Dexuan He , Xiao Zhou , Wenbin Guan , Liyuan Zhang , Xiaoman Zhang , Sinuo Xu , Ge Wang , Lifeng Wang
show 6 more authors
Xiaojun Yuan Xin Sun Yanfeng Wang Kun Sun Ya Zhang Weidi Xie
This is my paper
classification cs.CV
keywords rare cancer subtypingfew-shot learningvision-language foundation modelsprompt tuningmultiple instance learningwhole slide imagestile-level supervisiontumor region grounding
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

PathPT claims that vision-language pathology foundation models can be adapted to rare cancer subtyping under extreme data scarcity by treating whole-slide labels as a source of tile-level training signals rather than as slide-level classification targets. The framework freezes the vision and text encoders, adds a spatially-aware aggregator and learnable textual prompts, and uses zero-shot tile predictions to generate pseudo-labels that keep only tiles predicted normal or matching the slide's subtype. On eight rare cancer datasets (56 subtypes, roughly 2,900 WSIs) and three common cancer datasets, PathPT with the KEEP backbone reports consistent gains over four MIL baselines, including 0.679 balanced accuracy on the 30-subtype EBRAINS benchmark at 10 shots per subtype, along with improved tumor-region grounding.

Core claim

The central claim is that the bottleneck for few-shot rare-cancer subtyping is not feature aggregation but semantic alignment: conventional MIL treats VL models as frozen feature extractors and trains attention over visual features only, while PathPT shows that end-to-end prompt tuning over the frozen text encoder, combined with spatially-aware visual context, converts coarse slide-level labels into reliable tile-level supervision. The paper demonstrates this by benchmarking four VL backbones and four MIL frameworks, finding that gains track zero-shot grounding quality of the backbone and that PathPT outperforms all MIL baselines when paired with KEEP, on both classification and segmentation

What carries the argument

Three components carry PathPT. (1) Spatially-aware visual aggregation: parallel 3x3/5x5/7x7 residual convolutions capture local tile interactions and a transformer self-attention layer captures global context, applied to spatially ordered tile features. (2) Task-adaptive prompt tuning: K learnable tokens per class are prepended to the frozen text encoder (a CoOp-style context prompt), so the model learns class-specific text embeddings aligned with histopathological semantics. (3) Tile-level supervision from slide labels: zero-shot VL predictions assign pseudo-labels; tiles whose prediction is normal or matches the WSI label are kept as training signal, and an unlabeled-tile loss pushes remai

Load-bearing premise

PathPT's training signal is only as good as the zero-shot tile-level predictions of the vision-language backbone; if the backbone cannot distinguish tumor from normal at tile level, the pseudo-labels mislead training and PathPT falls behind MIL baselines.

What would settle it

Find or construct a backbone whose zero-shot tile DICE is at chance but whose features still allow MIL gains; if PathPT with that backbone still exceeds MIL, the pseudo-label explanation fails. Conversely, the paper's own PLIP results on TCGA datasets already show the predicted failure mode, offering a controlled case: applying PathPT with perfect ground-truth tile labels should restore the gains if the mechanism is as stated.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If PathPT's claims hold, few-shot rare-cancer subtyping can be done with 10 or fewer annotated slides per subtype, which is clinically meaningful for pediatric and rare adult tumors where expert annotations are scarce.
  • The approach yields tile-level predictions at inference, so the same model outputs both subtype and a spatial tumor map, giving pathologists a directly checkable localization rather than an opaque slide-level score.
  • Parameter-efficient adaptation (only prompt tokens and a lightweight aggregator are trained) makes it feasible to re-adapt systems when new subtypes are added, since the encoders stay frozen.
  • The reported dependence on zero-shot grounding quality means the method's ceiling is set by the backbone's ability to separate tumor from normal tissue; improvements in VL alignment should translate directly into PathPT gains.

Where Pith is reading between the lines

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

  • Because the selective pseudo-labeling acts as a noise filter (keeping only normal or label-consistent tiles), PathPT could serve as a general recipe for weakly-supervised tile classification in any domain where a zero-shot CLIP-like model provides partial grounding—e.g., histology beyond cancer, or medical imaging with coarse labels.
  • The method's failure mode with PLIP suggests a testable extension: explicitly model pseudo-label confidence or use multiple prompt-sets to detect unreliable backbones, potentially making the framework robust regardless of zero-shot quality.
  • One could test whether the gains come mostly from the spatial module or the prompts by ablating each on a backbone with strong grounding; the paper's ablation on segmentation suggests the combination matters, but a classification-only ablation across backbones would sharpen this.
  • If KEEP's advantage is its disease-knowledge injection, then coupling PathPT with knowledge-augmented backbones may be a general booster for rare-disease tasks—an indirect consequence the paper hints at but does not claim.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. PathPT is a prompt-tuning framework for pathology foundation models that replaces slide-level MIL aggregation with tile-level classification, using frozen vision/text encoders, a spatial-aware aggregator, learned prompts, and pseudo-labels derived from zero-shot VL predictions filtered by WSI labels. The paper benchmarks PathPT against ABMIL, CLAM, TransMIL, and DGRMIL on eight rare-cancer and three common-cancer subtyping datasets with 1/5/10-shot settings, and on three segmentation datasets. The headline result is PathPT-KEEP reaching 0.679 balanced accuracy on EBRAINS 10-shot; the paper claims consistent superiority and improved tumor-region grounding. The manuscript includes code and detailed supplementary tables.

Significance. If the few-shot comparison is valid, the method is a useful adaptation recipe for VL pathology models in low-data settings: it preserves tile-level interpretability, is parameter-efficient, and ships code. The curated pediatric benchmarks and comparisons across four backbones/four MIL baselines are valuable. However, the paper's own tables show the method only consistently outperforms MIL with the KEEP backbone; with PLIP, MUSK, and CONCH it often underperforms. The grounding DICE numbers are computed on training-set annotations, which weakens the interpretability claim. These issues are fixable by re-analysis and tempering claims.

major comments (4)
  1. [Section 4.4, Manual Prompt Selection] The 200 prompt groups are ranked by WSI classification performance 'on the training dataset.' The protocol in Section 2.2 defines a 15-WSI-per-subtype training split from which 1/5/10 WSIs are sampled for few-shot training. If prompt selection uses the full 15-WSI set, PathPT's tile pseudo-labels are generated with prompt groups chosen using labels that the MIL baselines never see. The paper never states that selection is restricted to the k-shot subset. Please clarify and re-run if necessary; this is essential for the EBRAINS KEEP margin (Table S1: 0.679 vs 0.650) to be a valid few-shot comparison.
  2. [Section 2.2 and Supplementary Tables S1-S11] The claim of 'consistently superior performance' is contradicted by the paper's own results. On EBRAINS with PLIP, PathPT 10-shot is 0.251 vs TransMIL 0.488 and DGRMIL 0.491 (Table S1); with CONCH 10-shot PathPT 0.491 vs TransMIL/DGRMIL 0.621. On TCGA-THYM PLIP 10-shot PathPT 0.430 vs TransMIL 0.683 (Table S3). PathPT is best mainly with KEEP, and is sometimes competitive with CONCH. Please restrict the 'superior' claim to the configurations where it holds and report win/loss counts with uncertainty.
  3. [Sections 2.2/2.3, Figures 2c/3b] The grounding DICE scores are computed on pathologist annotations of 'each WSI in the few-shot training set.' This evaluates the model's fit to training data, not its ability to localize unseen tumors. Since interpretability and grounding are central contributions, report DICE on held-out WSIs (e.g., test split of classification benchmarks) or explicitly state these are training-set localization results; the current wording implies a generalization claim.
  4. [Section 2.5 vs Table S12] 'PathPT consistently improved segmentation performance as the number of training samples increased' is not monotonic in the supplementary data: e.g., KEEP/CAMELYON16 DICE is 0.591 (5-shot) then 0.560 (10-shot); CONCH/CAMELYON16 is 0.522 then 0.505; MUSK/CAMELYON16 is 0.504 then 0.459. Please either correct the claim or analyze the non-monotonicity.
minor comments (4)
  1. [Abstract/Introduction/Section 2.3] Abstract says 2,910 WSIs and Introduction says 2,920; also pediatric total text says 1,232 but Table S16 sums to 1,229. Please harmonize.
  2. [Section 4.4] The selection of 'the top-performing 100 groups' is a hyperparameter; clarify whether it is per-dataset or per-split and how it interacts with the 10 repeats.
  3. [Throughout] Typos: 'detialed' (Section 2.2), 'lanaguage' (Section 2.3), 'vision-lanaguage' (Section 2.3). Supplementary Figure S2 caption says 'Rare adult cancer subtyping results' but text refers to pediatric; likely caption error.
  4. [Main subtyping tables] No statistical comparison is reported for the main subtyping tables; the segmentation section uses paired t-tests (Figure 5c). Add variance-aware comparisons or caveats, especially since some differences between methods are small relative to the reported interquartile ranges.

Circularity Check

0 steps flagged

No significant circularity; self-training on zero-shot pseudo-labels is standard and evaluation uses held-out labels.

full rationale

PathPT's training employs tile-level pseudo-labels derived from zero-shot predictions of the same frozen VL backbone. This is self-training/self-labeling, not logical circularity: the target of evaluation is held-out WSI subtype accuracy (Sec 4.8) and pathologist-mask DICE for segmentation (Sec 2.5), neither of which is used to derive the pseudo-labels. The paper reports that PathPT underperforms MIL with weak backbones, indicating the pseudo-labels are not trivially forcing improvement. The use of KEEP, a prior paper by overlapping authors, is a self-citation, but the central claim is supported across CONCH, PLIP, MUSK and against external benchmarks, so the self-citation is not load-bearing. A caveat is that the manual prompt selection (Sec 4.4) ranks prompts on 'the training dataset' without specifying whether this is the k-shot sampled subset or the full 15-WSI training set; if the latter, the few-shot protocol is weakened, but this is not a circular derivation. No equation reduces to its inputs by construction.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The method adds no new physical entities; its effectiveness rests on the assumption that the frozen VL model's zero-shot semantics are trustworthy enough to generate tile-level pseudo-labels, which is the main unverified premise.

free parameters (3)
  • Number of learnable prompt tokens K = 32
    Set by hand for all tasks; not tuned per dataset (Section 4.7).
  • Loss weights (labeled:unlabeled:pseudo-label) = 1:0.5:0.1
    Chosen by hand; not justified by experiments (Section 4.7).
  • Selected prompt groups = 100 of 200
    Prompt groups are ranked by WSI classification performance on the training set and the top 100 are selected, which is a training-set model selection step (Section 4.4).
axioms (4)
  • domain assumption Zero-shot VL tile predictions are reliable enough to act as pseudo-labels
    Section 4.4; results show this holds for KEEP but fails for PLIP/MUSK/CONCH, so the method's success depends on it.
  • domain assumption Unlabeled tiles in a WSI of subtype i belong to either normal tissue or subtype i
    Equation 7, candidate loss; this is a strong assumption that discards the possibility of other subtypes or mixed tissue in the WSI.
  • domain assumption WSI-level labels are correct and the tile with the slide label is informative
    Section 4.4; pseudo-labels conflicting with the WSI label are discarded, so a wrong slide label would poison tile-level training.
  • domain assumption Spatial ordering of tiles encodes morphological context useful for subtyping
    Section 4.3; the spatial-aware module assumes local and global tile interactions matter beyond bag-of-instance aggregation.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Boosting Pathology Foundation Models via Few-shot Prompt-tuning for Rare Cancer Subtyping." pith.science (2026). https://pith.science/paper/CAESZVON

@misc{pith2026250815904,
  author       = {Pith},
  title        = {Pith review of: Boosting Pathology Foundation Models via Few-shot Prompt-tuning for Rare Cancer Subtyping},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CAESZVON}},
  note         = {Machine review of arXiv:2508.15904}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Rare cancers comprise 20-25% of all malignancies but face major diagnostic challenges due to limited expert availability-especially in pediatric oncology, where they represent over 70% of cases. While pathology vision-language (VL) foundation models show promising zero-shot capabilities for common cancer subtyping, their clinical performance for rare cancers remains limited. Existing multi-instance learning (MIL) methods rely only on visual features, overlooking cross-modal knowledge and compromising interpretability critical for rare cancer diagnosis. To address this limitation, we propose PathPT, a novel framework that fully exploits the potential of vision-language pathology foundation models through spatially-aware visual aggregation and task-specific prompt tuning. Unlike conventional MIL, PathPT converts WSI-level supervision into fine-grained tile-level guidance by leveraging the zero-shot capabilities of VL models, thereby preserving localization on cancerous regions and enabling cross-modal reasoning through prompts aligned with histopathological semantics. We benchmark PathPT on eight rare cancer datasets(four adult and four pediatric) spanning 56 subtypes and 2,910 WSIs, as well as three common cancer datasets, evaluating four state-of-the-art VL models and four MIL frameworks under three few-shot settings. Results show that PathPT consistently delivers superior performance, achieving substantial gains in subtyping accuracy and cancerous region grounding ability. This work advances AI-assisted diagnosis for rare cancers, offering a scalable solution for improving subtyping accuracy in settings with limited access to specialized expertise.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

41 extracted references · 24 canonical work pages · 1 internal anchor

  1. [1]

    Machine learning-driven histotype diagnosis of ovarian carcinoma: Insights from the ocean ai challenge.medRxiv, pages 2024–04, 2024

    Maryam Asadi-Aghbolaghi, Hossein Farahani, Allen Zhang, Ardalan Akbari, Sirim Kim, Ashley Chow, Sohier Dane, OCEAN Challenge Consortium, OTTA Consortium, David G Huntsman, et al. Machine learning-driven histotype diagnosis of ovarian carcinoma: Insights from the ocean ai challenge.medRxiv, pages 2024–04, 2024

    work page 2024

  2. [2]

    Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer.JAMA, 318(22):2199–2210, 2017

    Babak Ehteshami Bejnordi, Mitko Veta, Paul Johannes Van Diest, Bram Van Ginneken, Nico Karssemeijer, Geert Litjens, Jeroen AWM Van Der Laak, Meyke Hermsen, Quirine F Manson, Maschenka Balkenhol, et al. Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer.JAMA, 318(22):2199–2210, 2017

    work page 2017

  3. [3]

    Artificial intelligence for diagnosis and gleason grading of prostate cancer: the panda challenge.Nature Medicine, 28(1):154–163, 2022

    Wouter Bulten, Kimmo Kartasalo, Po-Hsuan Cameron Chen, Peter Ström, Hans Pinckaers, Kunal Nagpal, Yuannan Cai, David F Steiner, Hester Van Boven, Robert Vink, et al. Artificial intelligence for diagnosis and gleason grading of prostate cancer: the panda challenge.Nature Medicine, 28(1):154–163, 2022

    work page 2022

  4. [4]

    Recent progress in the treatment of cancer in children.CA: a cancer journal for clinicians, 71(4):315–332, 2021

    Erin Butler, Kathleen Ludwig, Holly L Pacenta, Laura J Klesse, Tanya C Watt, and Theodore W Laetsch. Recent progress in the treatment of cancer in children.CA: a cancer journal for clinicians, 71(4):315–332, 2021

    work page 2021

  5. [5]

    Emerging properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerging properties in self-supervised vision transformers. InProceedings of the IEEE/CVF international conference on computer vision, pages 9650–9660, 2021

    2021

  6. [6]

    Towards a general-purpose foundation model for computational pathology.Nature Medicine, 2024

    Richard J Chen, Tong Ding, Ming Y Lu, Drew FK Williamson, Guillaume Jaume, Bowen Chen, Andrew Zhang, Daniel Shao, Andrew H Song, Muhammad Shaban, et al. Towards a general-purpose foundation model for computational pathology.Nature Medicine, 2024

    2024

  7. [7]

    Improved baselines with momentum contrastive learning.arXiv preprint arXiv:2003.04297, 2020

    Xinlei Chen, Haoqi Fan, Ross Girshick, and Kaiming He. Improved baselines with momentum contrastive learning.arXiv preprint arXiv:2003.04297, 2020

    Pith/arXiv arXiv 2003

  8. [8]

    The burden of rare cancers in the united states

    Carol E DeSantis, Joan L Kramer, and Ahmedin Jemal. The burden of rare cancers in the united states. CA: a cancer journal for clinicians, 67(4):261–272, 2017

    work page 2017

  9. [9]

    Multimodal whole slide foundation model for pathology.arXiv preprint arXiv:2411.19666, 2024

    Tong Ding, Sophia J Wagner, Andrew H Song, Richard J Chen, Ming Y Lu, Andrew Zhang, Anurag J Vaidya, Guillaume Jaume, Muhammad Shaban, Ahrong Kim, et al. Multimodal whole slide foundation model for pathology.arXiv preprint arXiv:2411.19666, 2024

    Pith/arXiv arXiv 2024

  10. [10]

    Deep learning-based histotype diagnosis of ovarian carcinoma whole-slide pathology images.Modern Pathology, 35(12):1983–1990, 2022

    HosseinFarahani, JeffreyBoschman, DavidFarnell, AmiraliDarbandsari, AllenZhang, PouyaAhmadvand, Steven JM Jones, David Huntsman, Martin Köbel, C Blake Gilks, et al. Deep learning-based histotype diagnosis of ovarian carcinoma whole-slide pathology images.Modern Pathology, 35(12):1983–1990, 2022

    work page 1983

  11. [11]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, and Ross Girshick. Masked autoencoders are scalable vision learners. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16000–16009, 2022

    2022

  12. [12]

    A visual–language foundation model for pathology image analysis using medical twitter.Nature Medicine, 29(9):2307–2316, 2023

    Zhi Huang, Federico Bianchi, Mert Yuksekgonul, Thomas J Montine, and James Zou. A visual–language foundation model for pathology image analysis using medical twitter.Nature Medicine, 29(9):2307–2316, 2023

    2023

  13. [13]

    A comprehensive ai model development framework for consistent gleason grading.Communications Medicine, 4(1):84, 2024

    Xinmi Huo, Kok Haur Ong, Kah Weng Lau, Laurent Gole, David M Young, Char Loo Tan, Xiaohui Zhu, Chongchong Zhang, Yonghui Zhang, Longjie Li, et al. A comprehensive ai model development framework for consistent gleason grading.Communications Medicine, 4(1):84, 2024

    work page 2024

  14. [14]

    Quilt-1m: One million image-text pairs for histopathology.Advances in Neural Information Processing Systems, 36, 2024

    Wisdom Ikezogwo, Saygin Seyfioglu, Fatemeh Ghezloo, Dylan Geva, Fatwir Sheikh Mohammed, Pa- van Kumar Anand, Ranjay Krishna, and Linda Shapiro. Quilt-1m: One million image-text pairs for histopathology.Advances in Neural Information Processing Systems, 36, 2024

    work page 2024

  15. [15]

    Attention-based deep multiple instance learning

    Maximilian Ilse, Jakub M Tomczak, and Max Welling. Attention-based deep multiple instance learning. arXiv preprint arXiv:1802.04712, 2018

    Pith/arXiv arXiv 2018

  16. [16]

    A visual-language foundation model for computational pathology.Nature Medicine, 30(3):863–874, 2024

    Ming Y Lu, Bowen Chen, Drew FK Williamson, Richard J Chen, Ivy Liang, Tong Ding, Guillaume Jaume, Igor Odintsov, Long Phi Le, Georg Gerber, et al. A visual-language foundation model for computational pathology.Nature Medicine, 30(3):863–874, 2024. |14

    work page 2024

  17. [17]

    Visual language pretrained multiple instance zero-shot transfer for histopathology images

    Ming Y Lu, Bowen Chen, Andrew Zhang, Drew FK Williamson, Richard J Chen, Tong Ding, Long Phi Le, Yung-Sung Chuang, and Faisal Mahmood. Visual language pretrained multiple instance zero-shot transfer for histopathology images. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19764–19775, 2023

    work page 2023

  18. [18]

    Data-efficient and weakly supervised computational pathology on whole-slide images.Nature Biomedical Engineering, 5(6):555–570, 2021

    Ming Y Lu, Drew FK Williamson, Tiffany Y Chen, Richard J Chen, Matteo Barbieri, and Faisal Mahmood. Data-efficient and weakly supervised computational pathology on whole-slide images.Nature Biomedical Engineering, 5(6):555–570, 2021

    2021

  19. [19]

    Towards a generalizable pathology foundation model via unified knowledge distillation.arXiv preprint arXiv:2407.18449, 2024

    Jiabo Ma, Zhengrui Guo, Fengtao Zhou, Yihui Wang, Yingxue Xu, Jinbang Li, Fang Yan, Yu Cai, Zhengjie Zhu, Cheng Jin, et al. Towards a generalizable pathology foundation model via unified knowledge distillation.arXiv preprint arXiv:2407.18449, 2024

    Pith/arXiv arXiv 2024

  20. [20]

    Hibou: A family of foundational vision transformers for pathology.arXiv preprint arXiv:2406.05074, 2024

    Dmitry Nechaev, Alexey Pchelnikov, and Ekaterina Ivanova. Hibou: A family of foundational vision transformers for pathology.arXiv preprint arXiv:2406.05074, 2024

    Pith/arXiv arXiv 2024

  21. [21]

    Socioeconomic inequalities in cancer incidence and access to health services among children and adolescents in china: a cross-sectional study.The Lancet, 400(10357):1020–1032, 2022

    Xin Ni, Zhe Li, Xinping Li, Xiao Zhang, Guoliang Bai, Yingying Liu, Rongshou Zheng, Yawei Zhang, Xin Xu, Yuanhu Liu, et al. Socioeconomic inequalities in cancer incidence and access to health services among children and adolescents in china: a cross-sectional study.The Lancet, 400(10357):1020–1032, 2022

    work page 2022

  22. [22]

    Linhao Qu, Kexue Fu, Manning Wang, Zhijian Song, et al. The rise of ai language pathologists: Exploring two-level prompt learning for few-shot weakly-supervised whole slide image classification.Advances in Neural Information Processing Systems, 36, 2024

    work page 2024

  23. [23]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. InInternational Conference on Machine Learning, pages 8748–8763, 2021

    2021

  24. [24]

    The digital brain tumour atlas, an open histopathology resource.Scientific Data, 9(1):55, 2022

    Thomas Roetzer-Pejrimovsky, Anna-Christina Moser, Baran Atli, Clemens Christian Vogel, Petra A Mercea, Romana Prihoda, Ellen Gelpi, Christine Haberler, Romana Höftberger, Johannes A Hainfellner, et al. The digital brain tumour atlas, an open histopathology resource.Scientific Data, 9(1):55, 2022

    work page 2022

  25. [25]

    Prism: A multi-modal generative foundation model for slide-level histopathology.arXiv preprint arXiv:2405.10254, 2024

    George Shaikovski, Adam Casson, Kristen Severson, Eric Zimmermann, Yi Kan Wang, Jeremy D Kunz, Juan A Retamero, Gerard Oakley, David Klimstra, Christopher Kanan, et al. Prism: A multi-modal generative foundation model for slide-level histopathology.arXiv preprint arXiv:2405.10254, 2024

    Pith/arXiv arXiv 2024

  26. [26]

    Transmil: Transformer based correlated multiple instance learning for whole slide image classification.Advances in Neural Information Processing Systems, 34:2136–2147, 2021

    Zhuchen Shao, Hao Bian, Yang Chen, Yifeng Wang, Jian Zhang, Xiangyang Ji, et al. Transmil: Transformer based correlated multiple instance learning for whole slide image classification.Advances in Neural Information Processing Systems, 34:2136–2147, 2021

    work page 2021

  27. [27]

    ViLa-MIL: Dual-scale vision- language multiple instance learning for whole slide image classification

    Jiangbo Shi, Chen Li, Tieliang Gong, Yefeng Zheng, and Huazhu Fu. ViLa-MIL: Dual-scale vision- language multiple instance learning for whole slide image classification. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11248–11258, 2024

    work page 2024

  28. [28]

    Cpath-omni: A unified multimodal foundation model for patch and whole slide image analysis in computational pathology

    Yuxuan Sun, Yixuan Si, Chenglu Zhu, Xuan Gong, Kai Zhang, Pingyi Chen, Ye Zhang, Zhongyi Shui, Tao Lin, and Lin Yang. Cpath-omni: A unified multimodal foundation model for patch and whole slide image analysis in computational pathology. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 10360–10371, 2025

    work page 2025

  29. [29]

    A foundation model for clinical-grade computational pathology and rare cancers detection.Nature Medicine, pages 1–12, 2024

    Eugene Vorontsov, Alican Bozkurt, Adam Casson, George Shaikovski, Michal Zelechowski, Kristen Severson, Eric Zimmermann, James Hall, Neil Tenenholtz, Nicolo Fusi, et al. A foundation model for clinical-grade computational pathology and rare cancers detection.Nature Medicine, pages 1–12, 2024

    work page 2024

  30. [30]

    Transformer-based unsupervised contrastive learning for histopathological image classification

    Xiyue Wang, Sen Yang, Jun Zhang, Minghui Wang, Jing Zhang, Wei Yang, Junzhou Huang, and Xiao Han. Transformer-based unsupervised contrastive learning for histopathological image classification. Medical Image Analysis, 81:102559, 2022

    work page 2022

  31. [31]

    A vision–language foundation model for precision oncology.Nature, pages 1–10, 2025

    Jinxi Xiang, Xiyue Wang, Xiaoming Zhang, Yinghua Xi, Feyisope Eweje, Yijiang Chen, Yuchen Li, Colin Bergstrom, Matthew Gopaulchan, Ted Kim, et al. A vision–language foundation model for precision oncology.Nature, pages 1–10, 2025

    work page 2025

  32. [32]

    A whole-slide foundation model for digital pathology from real-world data.Nature, pages 1–8, 2024

    Hanwen Xu, Naoto Usuyama, Jaspreet Bagga, Sheng Zhang, Rajesh Rao, Tristan Naumann, Cliff Wong, Zelalem Gero, Javier González, Yu Gu, et al. A whole-slide foundation model for digital pathology from real-world data.Nature, pages 1–8, 2024. |15

    work page 2024

  33. [33]

    A multimodal knowledge-enhanced whole-slide pathology foundation model

    Yingxue Xu, Yihui Wang, Fengtao Zhou, Jiabo Ma, Shu Yang, Huangjing Lin, Xin Wang, Jiguang Wang, Li Liang, Anjia Han, et al. A multimodal knowledge-enhanced whole-slide pathology foundation model. arXiv preprint arXiv:2407.15362, 2024

    Pith/arXiv arXiv 2024

  34. [34]

    A foundation model for generalizable cancer diagnosis and survival prediction from histopathological images.Nature Communications, 16(1):2366, 2025

    Zhaochang Yang, Ting Wei, Ying Liang, Xin Yuan, Ruitian Gao, Yujia Xia, Jie Zhou, Yue Zhang, and Zhangsheng Yu. A foundation model for generalizable cancer diagnosis and survival prediction from histopathological images.Nature Communications, 16(1):2366, 2025

    work page 2025

  35. [35]

    Coca: Contrastive captioners are image-text foundation models.arXiv preprint arXiv:2205.01917, 2022

    Jiahui Yu, Zirui Wang, Vijay Vasudevan, Legg Yeung, Mojtaba Seyedhosseini, and Yonghui Wu. Coca: Contrastive captioners are image-text foundation models.arXiv preprint arXiv:2205.01917, 2022

    Pith/arXiv arXiv 2022

  36. [36]

    Sigmoid loss for language image pre-training

    Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, and Lucas Beyer. Sigmoid loss for language image pre-training. InProceedings of the IEEE/CVF international conference on computer vision, pages 11975–11986, 2023

    2023

  37. [37]

    ibot: Image bert pre-training with online tokenizer.arXiv preprint arXiv:2111.07832, 2021

    Jinghao Zhou, Chen Wei, Huiyu Wang, Wei Shen, Cihang Xie, Alan Yuille, and Tao Kong. ibot: Image bert pre-training with online tokenizer.arXiv preprint arXiv:2111.07832, 2021

    Pith/arXiv arXiv 2021

  38. [38]

    Learning to prompt for vision-language models.International Journal of Computer Vision (IJCV), 2022

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. Learning to prompt for vision-language models.International Journal of Computer Vision (IJCV), 2022

    2022

  39. [39]

    A knowledge-enhanced pathology vision-language foundation model for cancer diagnosis.arXiv preprint arXiv:2412.13126, 2024

    Xiao Zhou, Luoyi Sun, Dexuan He, Wenbin Guan, Ruifen Wang, Lifeng Wang, Xin Sun, Kun Sun, Ya Zhang, Yanfeng Wang, et al. A knowledge-enhanced pathology vision-language foundation model for cancer diagnosis.arXiv preprint arXiv:2412.13126, 2024

    work page arXiv 2024

  40. [40]

    Knowledge-enhanced visual-language pretraining for computational pathology

    Xiao Zhou, Xiaoman Zhang, Chaoyi Wu, Ya Zhang, Weidi Xie, and Yanfeng Wang. Knowledge-enhanced visual-language pretraining for computational pathology. InEuropean Conference on Computer Vision, pages 345–362. Springer, 2024

    work page 2024

  41. [41]

    DGR-MIL: Exploring Diverse Global Representation in Multiple Instance Learning for Whole Slide Image Classification

    Wenhui Zhu, Xiwen Chen, Peijie Qiu, Aristeidis Sotiras, Abolfazl Razi, and Yalin Wang. Dgr-mil: Exploring diverse global representation in multiple instance learning for whole slide image classification. arXiv preprint arXiv:2407.03575, 2024. |16 Random Zero-shot ABMIL CLAM TransMIL DGRMIL PathPTRandom Zero-shot ABMIL CLAM TransMIL DGRMIL PathPT EBRAINS T...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.