Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

From Pixels to Gigapixels: Bridging Local Inductive Bias and Long-Range Dependencies with Pixel-Mamba

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Pixel-Mamba claims gigapixel whole-slide analysis can match pathology foundation models without any pathology-specific pretraining.

desk verdict Genuinely new end-to-end SSM architecture for gigapixel WSIs, but the headline claim of surpassing SOTA foundation models is not supported by the current evaluation. read the letter →

arxiv 2412.16711 v1 pith:CT7N64AU submitted 2024-12-21 cs.CV

Zhongwei Qiu , Hanqing Chao , Tiancheng Lin , Wanxing Chang , Zijiang Yang , Wenpei Jiao , Yixuan Shen , Yunshuo Zhang
show 7 more authors
Yelin Yang Wenbin Liu Hui Jiang Yun Bian Ke Yan Dakai Jin Le Lu
This is my paper · ORCID
classification cs.CV
keywords wholeslideimagesstatespacemodelMambaend-to-endtraininghierarchicalrepresentationcomputationalpathologytumorstagingsurvivalanalysis
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

Pixel-Mamba is an end-to-end architecture that consumes whole slide images at the pixel level and builds slide-level representations with Mamba, a state-space model whose memory cost grows linearly with sequence length. The paper's central claim is that by progressively expanding token receptive fields from 1x1 to 32x32 and fusing similar regions, the model captures both local and global hierarchy in gigapixel images. On tumor staging and survival analysis across three TCGA cancer datasets, Pixel-Mamba matches or beats two-stage MIL pipelines that rely on large pathology foundation models, even though Pixel-Mamba was pretrained only on ImageNet. If correct, this means pathology-specific pretraining at massive scale is not required for competitive slide-level analysis, and end-to-end training on raw pixels is a practical baseline.

What carries the argument

The central mechanism is progressive token expansion combined with region fusion inside a Mamba state-space backbone. Pixel tokens are serialized with a region-based zigzag scan, then each layer runs a bidirectional Mamba block to model long-range dependencies, while region fusion merges the most similar regions using CLS-token cosine similarity to cut redundancy and memory. Token expansion alternately concatenates and averages neighboring tokens along horizontal and vertical axes, growing the receptive field from 1x1 to 32x32 across 24 layers, so the network builds hierarchical multi-scale representations while keeping end-to-end training feasible on gigapixel inputs.

What would settle it

Reproduce the paper's protocol but select alpha, scan window, and magnification on held-out validation for each task separately, and run the two-stage baselines with their original published hyperparameters; if Pixel-Mamba's margins vanish or reverse, the claim that it matches foundation models without pathology pretraining fails.

Watch

Extended reading notes

Core claim

The paper claims that an end-to-end state-space model can outperform two-stage multiple-instance-learning pipelines that use pathology foundation models, without any pathology-specific pretraining. Concretely, Pixel-Mamba-Stage achieves macro F1 scores of 0.5334, 0.3744, and 0.3917 on BLCA, BRCA, and LUAD tumor staging, while Pixel-Mamba-Surv achieves C-index values of 0.6507, 0.6707, and 0.6468 on survival analysis for the same datasets. These numbers exceed the reported results of two-stage MIL with GigaPath, CONCH, and HIPT features, as well as the end-to-end LongViT baseline, with a 6.2M-parameter backbone.

Load-bearing premise

The central comparison is fair only if the reimplemented two-stage baselines were configured as carefully as Pixel-Mamba; the authors select their own alpha and scan window from the BLCA validation set, so if the baselines were not tuned similarly, the reported margin could shrink or disappear.

Editorial extensions

If this is right

  • If the claim holds, end-to-end state-space models become a competitive baseline for whole-slide analysis without needing expensive pathology foundation-model pretraining.
  • The linear memory scaling of Mamba could make end-to-end training on larger magnifications, such as 10x or 20x, tractable with modest GPU resources.
  • Progressive token expansion offers a general recipe for injecting locality into long-sequence vision models, potentially benefiting other gigapixel or high-resolution imaging tasks.
  • Region fusion based on CLS-token similarity suggests that redundancy reduction can be learned dynamically rather than fixed by patch grids.

Reading between the lines

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

  • The comparison may be sensitive to tuning asymmetries: the paper selects its own region-retention alpha and scan window from BLCA validation results, so an independent evaluation with per-task validation selection would clarify whether the margin reflects architecture or hyperparameter choice.
  • If the end-to-end advantage generalizes, it implies that spurious correlations in patch-level features, which two-stage pipelines inherit from pretrained encoders, can be corrected by jointly optimizing the whole slide representation.
  • A natural extension is to test Pixel-Mamba on molecular marker prediction or cancer subtyping tasks, where foundation-model embeddings currently dominate, to see whether raw-pixel end-to-end learning transfers beyond staging and survival.
  • The hierarchical token schedule from 1x1 to 32x32 could be transferred to other dense prediction domains, such as satellite or aerial imagery, where similar multi-scale local-global tradeoffs arise.
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 / 4 minor

Summary. The paper proposes Pixel-Mamba, an end-to-end state-space model for gigapixel whole-slide image analysis. Raw pixel tokens are serialized with a region-based zigzag scan; a bidirectional Mamba block provides global context; token expansion progressively enlarges receptive fields from 1x1 to 32x32; and a region-fusion module merges similar regions via CLS-token cosine similarity. The authors report ImageNet-1K classification and TCGA tumor staging and survival results, claiming that Pixel-Mamba, pretrained only on ImageNet, matches or exceeds foundation-model-based two-stage MIL pipelines without pathology-specific pretraining.

Significance. If the central comparison were fully supported, Pixel-Mamba would be a valuable lightweight end-to-end baseline for computational pathology, with linear-memory sequence modeling and a clear hierarchical design. The architecture is described in detail, and the ablations (Tables 3-6 and Table 9) usefully isolate the contributions of hierarchical tokenization, scan window, magnification, and the retention rate alpha. The ImageNet-1K results also show that Pixel-Mamba is competitive with existing SSM backbones at modest parameter counts. However, the main empirical claim of surpassing state-of-the-art pathology foundation models depends on benchmark comparisons that are currently undermined by missing baselines, in-sample hyperparameter selection, and overlapping error bars.

major comments (4)
  1. [Abstract and §4.3, Table 2] The central claim that Pixel-Mamba 'achieves or surpasses' SOTA pathology foundation models is not supported because the strongest contemporary foundation models, UNI and CHIEF, are absent from Table 2 even though both are cited in §2.2 as typical pathology FMs. GigaPath, CONCH, and HIPT are strong baselines, but they do not by themselves establish that the comparison is against SOTA. The authors should add UNI and CHIEF results or temper the abstract claim to say 'the evaluated foundation models.'
  2. [Appendix A.3, Table 9, and §4.5, Table 5] The flagship BLCA result is partly a fitted result. Appendix A.3 selects alpha = 0.8 by maximizing the C-index on BLCA, and Table 5 selects the scan-window size of 224 by evaluating on the same BLCA benchmark. These hyperparameters are then used for all downstream experiments, so the reported BLCA numbers are not an independent evaluation of the configuration. The authors should either use a held-out validation set or nested cross-validation for hyperparameter selection, or explicitly report the selection procedure and its impact on the main claims.
  3. [Table 2 and §4.3/§4.4] The reported advantages over two-stage baselines are often within one standard deviation, making the claim of 'outperforming' statistically unsupported. For example, BLCA survival C-index is 0.6507±0.0485 for Pixel-Mamba vs 0.6404±0.0253 for CONCH+TransMIL; BLCA staging macro-F1 is 0.5334±0.0608 vs 0.5320±0.0487 for GigaPath+ILRA-MIL. The paper provides no significance tests across folds or patient-level bootstrap intervals. The authors should add appropriate statistical tests and discuss whether these differences are meaningful.
  4. [§4.1 and Table 2] The two-stage baseline results need more implementation detail to be verifiable and fair. The text does not state the patch size, tissue-detection preprocessing, magnification, or whether official pretrained weights and feature-extraction code were used for GigaPath, CONCH, and HIPT. Since these FMs are known to be sensitive to patch-extraction protocol, the authors should document the exact feature-extraction pipeline and, if possible, release the extracted features or use official embeddings.
minor comments (4)
  1. [Eq. (1), §3.3] The notation in Eq. (1) is inconsistent: f_c is written as 'f c' in the text and the subscripts f_cx and f_cz are not defined. Please clarify which linear layers are shared and which are distinct, and define norm.
  2. [§4.3] The text refers to 'GigaViT' in the staging-results paragraph, but the table and related work call it GigaPath. Please use consistent names.
  3. [Table 2 and §4.1] The LongViT rows are labeled 0.6x, 2.5x, and 5.0x without explaining what these scaling factors mean in the main text. Please define them or refer to the supplementary material.
  4. [§3.2] The serialization step says each pixel is tokenized as an RGB token, but the input normalization (e.g., whether pixel values are scaled to [0,1]) is not described in the implementation details. Please specify the input preprocessing.

Circularity Check

1 steps flagged · score 2.0 of 10

No derivation-level circularity; low score reflects mild test-set hyperparameter selection on the BLCA benchmark used for the headline SOTA claim.

  1. fitted input called prediction [Appendix A.3 (Table 9); Section 4.5 Tables 5 and 6]
    "Pixel-Mamba-Surv achieves the best C-index of 0.6507 with the α = 0.8 on the BLCA dataset. Thus, we suggest α = 0.8, and all results of experiments in the main text of the manuscript are obtained with α = 0.8."

    The headline BLCA survival C-index (0.6507) is the same objective used to select α in the Region Fusion module, and Tables 5-6 show that the scan window (224) and magnification (2.5x) were likewise chosen by C-index on BLCA. The main-text claim that Pixel-Mamba outperforms all two-stage MIL methods on BLCA is therefore an optimized configuration point rather than an independent evaluation, so the reported BLCA comparison is partly forced by the selection procedure. This is a mild overfitting/test-selection circularity, not a derivation-level equivalence, and it does not affect the ImageNet pretraining results or the BRCA/LUAD comparisons in the same way.

full rationale

Pixel-Mamba is an empirical architecture paper: there is no first-principles derivation chain whose output could coincide with its inputs. The Mamba block is adopted from prior external work (Vim [54]), the scan-window idea from Local Mamba [16], and the survival loss from [52]; none of these are self-citations carrying a load-bearing uniqueness claim. The central comparison is against external TCGA benchmarks and ImageNet pretraining, so the architecture's content is independently evaluable. The only self-referential element is that α=0.8, the 224 scan window, and the 2.5x magnification were selected using the same BLCA C-index that is later reported as the headline result (Appendix A.3, Tables 5 and 6). This is test-set hyperparameter selection rather than a construction-level circularity: no equation reduces to its own input, and the BRCA/LUAD results and ImageNet classification are not fitted in this way. The SOTA-superiority claim is better criticized on the grounds of omitted baselines (UNI, CHIEF) and error-bar overlap, which are correctness risks rather than circularity.

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

The central empirical claim depends on several hand-chosen hyperparameters (alpha, scan window, input magnification, token expansion schedule) and on the transferability of ImageNet pretraining. No new physical entities are introduced.

free parameters (4)
  • Region fusion retention rate alpha = 0.8
    Selected by C-index ablation on BLCA (Appendix A.3, Table 9); controls how many regions are merged per layer and affects all reported results.
  • Scan window size = 224x224
    Set to 224 because pretraining used 224x224 images (Section 4.5, Table 5); determines sequence locality and is a key design choice.
  • Token expansion schedule = Hand-designed per Tables 7 and 8
    The layer-wise schedule of horizontal and vertical expansion is set by hand, not learned or searched, yet is central to the hierarchical representation claim.
  • Input magnification = 2.5x
    Downstream fine-tuning uses 2.5x because of GPU memory limits (Section 4.5, Table 6); scaling results depend on this resolution.
assumptions (4)
  • domain assumption Mamba and SSM blocks capture long-range dependencies in ultra-long token sequences with linear memory complexity.
    Invoked in Sections 1 and 3.3 as the core efficiency mechanism; this is a property of the Mamba module, not shown for this architecture.
  • domain assumption ImageNet pretraining transfers to histopathology sufficiently for the architecture to outperform pathology-specific pretraining.
    The main claim depends on initialization from ImageNet classification (Section 4.1, A.2) rather than any pathology data.
  • ad hoc to paper Regions whose CLS tokens are similar by cosine similarity are redundant and can be averaged without losing predictive information.
    Region Fusion (Section 3.3) discards information by averaging matched regions; the paper does not test whether this assumption holds for tumor staging or survival.
  • ad hoc to paper Zigzag region-based serialization preserves enough spatial locality for local inductive biases to emerge.
    The serialization design (Section 3.2) assumes that arranging pixels in h x w windows and concatenating regions captures neighborhood structure; no analysis of alternative orderings beyond scan-window size is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Pixels to Gigapixels: Bridging Local Inductive Bias and Long-Range Dependencies with Pixel-Mamba." pith.science (2026). https://pith.science/paper/CT7N64AU

@misc{pith2026241216711,
  author       = {Pith},
  title        = {Pith review of: From Pixels to Gigapixels: Bridging Local Inductive Bias and Long-Range Dependencies with Pixel-Mamba},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CT7N64AU}},
  note         = {Machine review of arXiv:2412.16711}
}
read the original abstract

Histopathology plays a critical role in medical diagnostics, with whole slide images (WSIs) offering valuable insights that directly influence clinical decision-making. However, the large size and complexity of WSIs may pose significant challenges for deep learning models, in both computational efficiency and effective representation learning. In this work, we introduce Pixel-Mamba, a novel deep learning architecture designed to efficiently handle gigapixel WSIs. Pixel-Mamba leverages the Mamba module, a state-space model (SSM) with linear memory complexity, and incorporates local inductive biases through progressively expanding tokens, akin to convolutional neural networks. This enables Pixel-Mamba to hierarchically combine both local and global information while efficiently addressing computational challenges. Remarkably, Pixel-Mamba achieves or even surpasses the quantitative performance of state-of-the-art (SOTA) foundation models that were pretrained on millions of WSIs or WSI-text pairs, in a range of tumor staging and survival analysis tasks, {\bf even without requiring any pathology-specific pretraining}. Extensive experiments demonstrate the efficacy of Pixel-Mamba as a powerful and efficient framework for end-to-end WSI analysis.

Figures

Figures reproduced from arXiv: 2412.16711 by the authors.

Figure 1
Figure 1. (a) Pathologists integrate observations from multiple re [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The Pixel-Mamba Framework. (a) The WSI is serialized, with CLS tokens added to create the token series [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The illustration of Token Expansion in a region. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The comparison of Kaplan-Meier analysis and Log [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Atlas: A Novel Pathology Foundation Model by Mayo Clinic, Charit\'e, and Aignostics

    cs.CV 2025-01 conditional novelty 5.0 of 10

    Atlas, a 632M-parameter ViT pathology model trained on 1.2M multi-stain slides, achieves a 61.9 percent average on 21 public benchmarks, the best among seven leading foundation models.

Reference graph

Works this paper leans on

54 extracted references · 43 canonical work pages · cited by 1 Pith paper

  1. [1]

    An image is worth 16x16 words: Trans- formers for image recognition at scale

    Dosovitskiy Alexey. An image is worth 16x16 words: Trans- formers for image recognition at scale. ICLR, 2021. 2, 3, 5, 6

    work page 2021

  2. [2]

    Tertiary lymphoid structures (tls) identification and density assessment on h&e- stained digital slides of lung cancer

    Panagiotis Barmpoutis, Matthew Di Capite, Hamzeh Kay- hanian, William Waddingham, Daniel C Alexander, Marnix Jansen, and Francois Ng Kee Kwong. Tertiary lymphoid structures (tls) identification and density assessment on h&e- stained digital slides of lung cancer. Plos one , 16(9): e0256907, 2021. 2

    work page 2021

  3. [3]

    Scaling vision transformers to gigapixel images via hierarchical self-supervised learning

    Richard J Chen, Chengkuan Chen, Yicong Li, Tiffany Y Chen, Andrew D Trister, Rahul G Krishnan, and Faisal Mahmood. Scaling vision transformers to gigapixel images via hierarchical self-supervised learning. In CVPR, pages 16144–16155, 2022. 3

    work page 2022

  4. [4]

    Scaling vision transformers to gigapixel images via hierarchical self-supervised learning

    Richard J Chen, Chengkuan Chen, Yicong Li, Tiffany Y Chen, Andrew D Trister, Rahul G Krishnan, and Faisal Mahmood. Scaling vision transformers to gigapixel images via hierarchical self-supervised learning. In CVPR, pages 16144–16155, 2022. 2, 6, 7

    work page 2022

  5. [5]

    Towards a general-purpose foundation model for com- putational pathology

    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 com- putational pathology. Nature Medicine, 2024. 2, 3, 8

    work page 2024

  6. [6]

    Diagnose like a pathologist: Weakly-supervised pathologist-tree network for slide-level immunohistochemical scoring

    Zhen Chen, Jun Zhang, Shuanlong Che, Junzhou Huang, Xiao Han, and Yixuan Yuan. Diagnose like a pathologist: Weakly-supervised pathologist-tree network for slide-level immunohistochemical scoring. In AAAI, pages 47–54, 2021. 3

    work page 2021

  7. [7]

    Spatial architecture and arrangement of tumor-infiltrating lympho- cytes for predicting likelihood of recurrence in early-stage non–small cell lung cancer

    Germ ´an Corredor, Xiangxue Wang, Yu Zhou, Cheng Lu, Pingfu Fu, Konstantinos Syrigos, David L Rimm, Michael Yang, Eduardo Romero, Kurt A Schalper, et al. Spatial architecture and arrangement of tumor-infiltrating lympho- cytes for predicting likelihood of recurrence in early-stage non–small cell lung cancer. Clinical cancer research, 25(5): 1526–1534, 2019. 1

    work page 2019

  8. [8]

    Transformers are ssms: General- ized models and efficient algorithms through structured state space duality

    Tri Dao and Albert Gu. Transformers are ssms: General- ized models and efficient algorithms through structured state space duality. arXiv preprint arXiv:2405.21060, 2024. 3

    arXiv 2024

Show all 54 references
  1. [9]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In CVPR, pages 248–255. Ieee, 2009. 6, 9

    2009

  2. [10]

    Longnet: Scaling transformers to 1,000,000,000 tokens

    Jiayu Ding, Shuming Ma, Li Dong, Xingxing Zhang, Shao- han Huang, Wenhui Wang, Nanning Zheng, and Furu Wei. Longnet: Scaling transformers to 1,000,000,000 tokens. arXiv preprint arXiv:2307.02486, 2023. 3, 6

    2023 arXiv

  3. [11]

    Structured state space models for multiple instance learning in digital pathol- ogy

    Leo Fillioux, Joseph Boyd, Maria Vakalopoulou, Paul-Henry Courn`ede, and Stergios Christodoulidis. Structured state space models for multiple instance learning in digital pathol- ogy. In MICCAI, pages 594–604. Springer, 2023. 2, 3

    2023

  4. [12]

    Mamba: Linear-time sequence modeling with selective state spaces

    Albert Gu and Tri Dao. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752, 2023. 2, 3

    2023 arXiv

  5. [13]

    Multi-scale domain-adversarial multiple-instance cnn for cancer subtype classification with unannotated histopatho- logical images

    Noriaki Hashimoto, Daisuke Fukushima, Ryoichi Koga, Yusuke Takagi, Kaho Ko, Kei Kohno, Masato Nakaguro, Shigeo Nakamura, Hidekata Hontani, and Ichiro Takeuchi. Multi-scale domain-adversarial multiple-instance cnn for cancer subtype classification with unannotated histopatho- l...

    2020

  6. [14]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR, pages 770–778, 2016. 2, 3, 5, 6, 7

    2016

  7. [15]

    Hˆ 2-mil: ex- ploring hierarchical representation with heterogeneous mul- tiple instance learning for whole slide image analysis

    Wentai Hou, Lequan Yu, Chengxuan Lin, Helong Huang, Rongshan Yu, Jing Qin, and Liansheng Wang. Hˆ 2-mil: ex- ploring hierarchical representation with heterogeneous mul- tiple instance learning for whole slide image analysis. In AAAI, pages 933–941, 2022. 3

    2022

  8. [16]

    Localmamba: Visual state space model with windowed selective scan

    Tao Huang, Xiaohuan Pei, Shan You, Fei Wang, Chen Qian, and Chang Xu. Localmamba: Visual state space model with windowed selective scan. arXiv preprint arXiv:2403.09338,

    arXiv

  9. [17]

    Integration of patch features through self-supervised learning and transformer for survival analysis on whole slide images

    Ziwang Huang, Hua Chai, Ruoqi Wang, Haitao Wang, Yue- dong Yang, and Hejun Wu. Integration of patch features through self-supervised learning and transformer for survival analysis on whole slide images. In MICCAI, pages 561–570. Springer, 2021. 3

    2021

  10. [18]

    A visual–language foundation model for pathology image analysis using medical twitter

    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. 3

    2023

  11. [19]

    Attention-based deep multiple instance learning

    Maximilian Ilse, Jakub Tomczak, and Max Welling. Attention-based deep multiple instance learning. In ICML, pages 2127–2136. PMLR, 2018. 2, 3, 6, 7

    2018

  12. [20]

    Hest-1k: A dataset for spatial transcriptomics and histology image analysis

    Guillaume Jaume, Paul Doucet, Andrew H Song, Ming Y Lu, Cristina Almagro-P ´erez, Sophia J Wagner, Anurag J Vaidya, Richard J Chen, Drew FK Williamson, Ahrong Kim, et al. Hest-1k: A dataset for spatial transcriptomics and histology image analysis. arXiv preprint arXiv:2406.16192,

    arXiv

  13. [21]

    Self-path: Self-supervision for classification of pathology images with limited annotations

    Navid Alemi Koohbanani, Balagopal Unnikrishnan, Syed Ali Khurram, Pavitra Krishnaswamy, and Nasir Rajpoot. Self-path: Self-supervision for classification of pathology images with limited annotations. TMI, 40(10): 2845–2856, 2021. 3

    2021

  14. [22]

    Dual-stream multi- ple instance learning network for whole slide image classi- fication with self-supervised contrastive learning

    Bin Li, Yin Li, and Kevin W Eliceiri. Dual-stream multi- ple instance learning network for whole slide image classi- fication with self-supervised contrastive learning. In CVPR, pages 14318–14328, 2021. 2, 3

    2021

  15. [23]

    Dynamic graph repre- sentation with knowledge-aware attention for histopathology whole slide image analysis

    Jiawen Li, Yuxuan Chen, Hongbo Chu, Qiehe Sun, Tian Guan, Anjia Han, and Yonghong He. Dynamic graph repre- sentation with knowledge-aware attention for histopathology whole slide image analysis. In CVPR, pages 11323–11332,

  16. [24]

    Vmamba: Visual state space model

    Yue Liu, Yunjie Tian, Yuzhong Zhao, Hongtian Yu, Lingxi Xie, Yaowei Wang, Qixiang Ye, and Yunfan Liu. Vmamba: Visual state space model. arXiv preprint arXiv:2401.10166,

    arXiv

  17. [25]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In ICCV, pages 10012–10022, 2021. 3

    2021

  18. [26]

    Data-efficient and weakly supervised computational pathology on whole- slide images

    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. 2, 3

    2021

  19. [27]

    A visual- language foundation model for computational pathology

    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. 2, 3, 6, 7

    2024

  20. [28]

    Self-distillation augmented masked autoen- coders for histopathological image understanding

    Yang Luo, Zhineng Chen, Shengtian Zhou, Kai Hu, and Xieping Gao. Self-distillation augmented masked autoen- coders for histopathological image understanding. In BIBM, pages 1343–1349. IEEE, 2023. 3

    2023

  21. [29]

    Vim4path: Self-supervised vision mamba for histopathology images

    Ali Nasiri-Sarvi, Vincent Quoc-Huy Trinh, Hassan Rivaz, and Mahdi S Hosseini. Vim4path: Self-supervised vision mamba for histopathology images. In CVPR, pages 6894– 6903, 2024. 3

    2024

  22. [30]

    S4nd: Modeling images and videos as multidimensional signals with state spaces

    Eric Nguyen, Karan Goel, Albert Gu, Gordon Downs, Preey Shah, Tri Dao, Stephen Baccus, and Christopher R ´e. S4nd: Modeling images and videos as multidimensional signals with state spaces. NeurIPS, 35:2846–2861, 2022. 6

    2022

  23. [31]

    Ivt: An end-to-end instance-guided video transformer for 3d pose estimation

    Zhongwei Qiu, Qiansheng Yang, Jian Wang, and Dongmei Fu. Ivt: An end-to-end instance-guided video transformer for 3d pose estimation. In ACM MM, pages 6174–6182, 2022. 2

    2022

  24. [32]

    Learning degradation-robust spatiotemporal frequency-transformer for video super- resolution

    Zhongwei Qiu, Huan Yang, Jianlong Fu, Daochang Liu, Chang Xu, and Dongmei Fu. Learning degradation-robust spatiotemporal frequency-transformer for video super- resolution. TPAMI, 2023. 2

    2023

  25. [33]

    End- to-end multi-source visual prompt tuning for survival analy- sis in whole slide images

    Zhongwei Qiu, Hanqing Chao, Wenbin Liu, Yixuan Shen, Le Lu, Ke Yan, Dakai Jin, Yun Bian, and Hui Jiang. End- to-end multi-source visual prompt tuning for survival analy- sis in whole slide images. arXiv preprint arXiv:2409.03804,

    arXiv

  26. [34]

    Global contrast-masked autoencoders are powerful pathological representation learners

    Hao Quan, Xingyu Li, Weixing Chen, Qun Bai, Mingchen Zou, Ruijie Yang, Tingting Zheng, Ruiqun Qi, Xinghua Gao, and Xiaoyu Cui. Global contrast-masked autoencoders are powerful pathological representation learners. PR, 156: 110745, 2024. 3

    2024

  27. [35]

    Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images

    Joel Saltz, Rajarsi Gupta, Le Hou, Tahsin Kurc, Pankaj Singh, Vu Nguyen, Dimitris Samaras, Kenneth R Shroyer, Tianhao Zhao, Rebecca Batiste, et al. Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images. Cell re...

    2018

  28. [36]

    Transmil: Transformer based correlated multiple instance learning for whole slide image classification

    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. NeurIPS, 34:2136–2147, 2021. 2, 6, 7

    2021

  29. [37]

    Adaptive weighting multi-field-of-view cnn for semantic segmentation in pathology

    Hiroki Tokunaga, Yuki Teramoto, Akihiko Yoshizawa, and Ryoma Bise. Adaptive weighting multi-field-of-view cnn for semantic segmentation in pathology. InCVPR, pages 12597– 12606, 2019. 3

    2019

  30. [38]

    Training data-efficient image transformers & distillation through at- tention

    Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Herv ´e J´egou. Training data-efficient image transformers & distillation through at- tention. In ICML, pages 10347–10357. PMLR, 2021. 6

    2021

  31. [39]

    Multi-resolution deep learning characterizes tertiary lymphoid structures and their prognostic relevance in solid tumors

    Mart van Rijthoven, Simon Obahor, Fabio Pagliarulo, Maries van den Broek, Peter Schraml, Holger Moch, Jeroen van der Laak, Francesco Ciompi, and Karina Silina. Multi-resolution deep learning characterizes tertiary lymphoid structures and their prognostic relevance in solid tum...

    2024

  32. [40]

    Rethinking multiple instance learning for whole slide image classification: A bag-level classifier is a good instance-level teacher

    Hongyi Wang, Luyang Luo, Fang Wang, Ruofeng Tong, Yen-Wei Chen, Hongjie Hu, Lanfen Lin, and Hao Chen. Rethinking multiple instance learning for whole slide image classification: A bag-level classifier is a good instance-level teacher. TMI, 2024. 3

    2024

  33. [41]

    When an image is worth 1,024 x 1,024 words: A case study in computational pathology

    Wenhui Wang, Shuming Ma, Hanwen Xu, Naoto Usuyama, Jiayu Ding, Hoifung Poon, and Furu Wei. When an image is worth 1,024 x 1,024 words: A case study in computational pathology. arXiv preprint arXiv:2312.03558, 2023. 2, 3, 6, 7, 8

    2023 arXiv

  34. [42]

    Transpath: Transformer-based self-supervised learning for histopatho- logical image classification

    Xiyue Wang, Sen Yang, Jun Zhang, Minghui Wang, Jing Zhang, Junzhou Huang, Wei Yang, and Xiao Han. Transpath: Transformer-based self-supervised learning for histopatho- logical image classification. In MICCAI, pages 186–195. Springer, 2021. 2, 3

    2021

  35. [43]

    A pathology foundation model for can- cer diagnosis and prognosis prediction

    Xiyue Wang, Junhan Zhao, Eliana Marostica, Wei Yuan, Ji- etian Jin, Jiayu Zhang, Ruijiang Li, Hongping Tang, Kanran Wang, Yu Li, et al. A pathology foundation model for can- cer diagnosis and prognosis prediction. Nature, pages 1–9,

  36. [44]

    Exploring low-rank property in multiple instance learning for whole slide image classifica- tion

    Jinxi Xiang and Jun Zhang. Exploring low-rank property in multiple instance learning for whole slide image classifica- tion. In ICLR, 2023. 2, 6, 7

    2023

  37. [45]

    Aggregated residual transformations for deep neural networks

    Saining Xie, Ross Girshick, Piotr Doll ´ar, Zhuowen Tu, and Kaiming He. Aggregated residual transformations for deep neural networks. In CVPR, pages 1492–1500, 2017. 6

    2017

  38. [46]

    Diagnose like a pathologist: Transformer-enabled hierarchical attention-guided multiple instance learning for whole slide image classification

    Conghao Xiong, Hao Chen, Joseph JY Sung, and Irwin King. Diagnose like a pathologist: Transformer-enabled hierarchical attention-guided multiple instance learning for whole slide image classification. arXiv preprint arXiv:2301.08125, 2023. 3

    2023 arXiv

  39. [47]

    A whole-slide foundation model for digital pathology from real-world data

    Hanwen Xu, Naoto Usuyama, Jaspreet Bagga, Sheng Zhang, Rajesh Rao, Tristan Naumann, Cliff Wong, Zelalem Gero, Javier Gonz ´alez, Yu Gu, et al. A whole-slide foundation model for digital pathology from real-world data. Nature, pages 1–8, 2024. 3

    2024

  40. [48]

    A whole-slide foundation model for digital pathology from real-world data

    Hanwen Xu, Naoto Usuyama, Jaspreet Bagga, Sheng Zhang, Rajesh Rao, Tristan Naumann, Cliff Wong, Zelalem Gero, Javier Gonz ´alez, Yu Gu, et al. A whole-slide foundation model for digital pathology from real-world data. Nature, pages 1–8, 2024. 2, 3, 6, 7, 8

    2024

  41. [49]

    Self-supervised visual representation learn- ing for histopathological images

    Pengshuai Yang, Zhiwei Hong, Xiaoxu Yin, Chengzhan Zhu, and Rui Jiang. Self-supervised visual representation learn- ing for histopathological images. In MICCAI, pages 47–57. Springer, 2021. 3

    2021

  42. [50]

    Mambamil: En- hancing long sequence modeling with sequence reordering in computational pathology

    Shu Yang, Yihui Wang, and Hao Chen. Mambamil: En- hancing long sequence modeling with sequence reordering in computational pathology. arXiv preprint arXiv:2403.06800,

    arXiv

  43. [51]

    Whole slide images based cancer survival prediction using attention guided deep mul- tiple instance learning networks

    Jiawen Yao, Xinliang Zhu, Jitendra Jonnagaddala, Nicholas Hawkins, and Junzhou Huang. Whole slide images based cancer survival prediction using attention guided deep mul- tiple instance learning networks. MIA, 65:101789, 2020. 2

    2020

  44. [52]

    Bias in cross- entropy-based training of deep survival networks

    Shekoufeh Gorgi Zadeh and Matthias Schmid. Bias in cross- entropy-based training of deep survival networks. TPAMI, 43(9):3126–3137, 2020. 5

    2020

  45. [53]

    Dtfd- mil: Double-tier feature distillation multiple instance learn- ing for histopathology whole slide image classification

    Hongrun Zhang, Yanda Meng, Yitian Zhao, Yihong Qiao, Xiaoyun Yang, Sarah E Coupland, and Yalin Zheng. Dtfd- mil: Double-tier feature distillation multiple instance learn- ing for histopathology whole slide image classification. In CVPR, pages 18802–18812, 2022. 2

    2022

  46. [54]

    Vision mamba: Efficient visual representation learning with bidirectional state space model

    Lianghui Zhu, Bencheng Liao, Qian Zhang, Xinlong Wang, Wenyu Liu, and Xinggang Wang. Vision mamba: Efficient visual representation learning with bidirectional state space model. arXiv preprint arXiv:2401.09417, 2024. 2, 3, 4, 5, 6, 8, 9

    2024 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.