Pith. sign in

REVIEW 5 major objections 6 minor 85 references

FMaMIL: Frequency-Driven Mamba Multi-Instance Learning for Weakly Supervised Lesion Segmentation in Medical Images

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read FMaMIL reports weakly supervised lesion segmentation from image-level labels alone, with mIoU/Dice of 0.887/0.934 on glomerular lesions and 0.869/0.957 on CAMELYON16.

desk verdict Useful architecture, untrustworthy CAMELYON16 protocol—needs clarification of patch-label usage and a clean held-out evaluation before the numbers mean anything. read the letter →

arxiv 2506.07652 v1 pith:UFUGD2MY submitted 2025-06-09 cs.CV cs.AI

Hangbei Cheng , Xiaorong Dong , Xueyu Liu , Jianan Zhang , Xuetao Ma , Mingqiang Wei , Liansheng Wang , Junxin Chen
show 1 more author
Yongfei Wu
This is my paper
classification cs.CVcs.AI
keywords Frequency-domainhistopathologyimagesmultipleinstancelearninglesionsegmentationMambamodel
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper aims to show that lesion segmentation in histopathology can be driven by image-level labels alone, without pixel-level annotations. The proposed FMaMIL pipeline first uses a Mamba-based multi-instance encoder, boosted by learnable Fourier-domain filtering and bidirectional row/column scanning, to produce class activation maps (CAMs); then a U-Net is trained with CAM-based soft supervision, binarized hard labels, and a confidence-weighted self-correction loss. On a private glomerular-lesion dataset and the public CAMELYON16 dataset, the paper reports mIoU/Dice of 0.887/0.934 and 0.869/0.957 respectively, above the weakly supervised and fully supervised baselines in Table 2. The intended payoff is that diagnostic histopathology segmentation could be trained where only case-level diagnoses are available.

What carries the argument

The load-bearing component is the FMamba block, a Mamba encoder layer with three parallel branches: row-scanned and column-scanned sequences, each processed by forward and backward selective state-space models and gated with a SiLU-activated branch, plus a learnable frequency-domain encoding that runs FFT along the sequence, applies learned complex frequency weights, runs IFFT, and adds a skip connection. The three branch outputs are summed and passed through a residual connection. This mechanism converts a sequence of image patches into a fused feature sequence in linear time, capturing long-range dependencies while injecting spectral texture and boundary information; the attention weights from the MIL classification head are then converted into class activation maps that seed second-stage segmentation training.

What would settle it

Train the identical pipeline on CAMELYON16 with true bag-level supervision, meaning no per-patch labels are assigned during preprocessing, and compare segmentation mIoU/Dice; if performance falls to the level of the ablated model without the frequency module, the reported gains came from the patch-label shortcut described in Section 4.1 rather than from the Mamba or frequency architecture.

Watch

Extended reading notes

Core claim

The central claim is that combining a selective state-space model with frequency-domain information makes a weak multi-instance learner produce substantially better class activation maps, and that those maps, when refined by soft supervision and self-correction, are accurate enough to supervise a segmentation model. The paper reports classification accuracy/AUC of 0.996/0.998 on the glomerular dataset and 0.993/0.992 on CAMELYON16, with segmentation mIoU/Dice of 0.887/0.934 and 0.869/0.957 respectively, and states that these numbers exceed all weakly supervised competitors and also the fully supervised U-Net and DeepLabV3+ in Table 2. The gain is attributed to three cooperating modules: a learnable frequency-domain encoding that applies FFT, multiplies the spectrum by learned complex weights, and returns via IFFT with a skip connection; a bidirectional scanning strategy that models inter-patch context in both row and column orders; and an attention-based instance-to-bag classification head whose instance-level loss sharpens the attention maps that become CAMs.

Load-bearing premise

The method's claim depends on the first stage learning from slide-level labels alone, but the paper says each patch in CAMELYON16 was assigned the label of its slide, so positive slides contribute many positively labeled patches; if that is what was done, the model may be doing ordinary supervised patch classification instead of multi-instance learning.

Editorial extensions

If this is right

  • Segmentation models for histopathology could be trained from case-level diagnoses alone in the evaluated settings, removing the need for pixel-level pathologist annotation.
  • On the paper's two test sets, the pipeline reports segmentation quality above fully supervised U-Net and DeepLabV3+, suggesting that weak supervision is not necessarily a ceiling relative to pixel-trained models when annotations are sparse or noisy.
  • Because the Mamba encoder is linear in sequence length, the approach scales to the long patch sequences of whole-slide images more efficiently than transformer-based multi-instance baselines.
  • The ablation attributes the largest single-stage gain to the learnable frequency-domain encoding, indicating spectral features carry much of the segmentation improvement independently of the scanning strategy.

Reading between the lines

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

  • Editorial inference: The CAMELYON16 preprocessing described in Section 4.1, where the label for each patch was assigned from the patient-level label, conflicts with the multi-instance premise that instance labels are unknown; if taken literally, the first stage could be learning from noisy per-patch labels inherited from positive slides, and the reported gains may not isolate the Mamba or frequenc
  • Editorial inference: The optimal patch size of 100x100 on 1100x1100 images is close to the scale of glomerular structures, so a natural testable extension is whether multi-scale or lesion-scale-normalized patch sampling preserves the gains on datasets with variable lesion sizes.
  • Editorial inference: The reported edge over fully supervised baselines may depend on incomplete pixel annotations in the evaluation sets; on a densely annotated dataset the fully supervised models might narrow the gap, so an evaluation with exhaustive annotations would sharpen the comparison.
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

5 major / 6 minor

Summary. The paper proposes FMaMIL, a two-stage weakly supervised framework for lesion segmentation in histopathology. Stage one extracts patch features with a pre-trained CNN, applies a Mamba-based encoder with bidirectional spatial scanning and a learnable frequency-domain encoding module, and trains an attention-based MIL head with both bag-level and instance-level losses. Stage two generates CAMs from this head, binarizes them into pseudo-labels, and trains a U-Net with a combination of soft and hard losses, plus a confidence-based pseudo-label refinement and consistency loss. The method is evaluated on a private glomerular-lesion dataset and on CAMELYON16, with claims of state-of-the-art weakly supervised segmentation performance and, surprisingly, performance above fully supervised U-Net and DeepLabV3+.

Significance. If the reported results are valid, the architectural combination—Mamba-based long-range modeling, learnable frequency encoding, and a two-stage pseudo-label refinement scheme—would be a useful contribution to weakly supervised histopathology segmentation, and the public code link would aid reproducibility. The core ideas are plausible and the problem is practically important. However, the current evidence is not sufficient to support the head-line claims: the CAMELYON16 preprocessing described in Section 4.1 conflicts with the MIL assumptions of Section 3.1.2, the instance-level loss in Eq. (23) is not well-defined, and the evaluation protocol tunes hyperparameters on the same validation set used for final comparisons. These issues are central to the empirical claims rather than superficial presentation defects.

major comments (5)
  1. [§4.1, §3.1.2] The CAMELYON16 description states that "the label for each patch was assigned based on the patient-level label," which directly contradicts the MIL formulation of Section 3.1.2 (Eq. (5)) where instance labels are latent and only bag labels are observed. If these per-patch labels are used in any training loss, such as the instance-level loss in Eq. (23), FMaMIL receives noisy dense per-patch supervision rather than bag-level supervision. That would make the comparison with bag-level baselines (AB-MIL, Trans-MIL, MambaMIL) invalid as a test of the proposed MIL or frequency modules. The paper must clarify exactly how patch labels are used in each stage, and if they are used only to construct bags, state this explicitly.
  2. [§3.5, Eq. (23)] Equation (23) defines L_instance = -Σ_i α_i log ŷ_i but contains no ground-truth or soft target for the instance. As written, minimizing this term drives every instance prediction toward 1, since α_i is nonnegative and log ŷ_i is maximized when ŷ_i = 1. If ŷ_i is a probability vector, the expression is vector-valued rather than a scalar loss. The instance-level supervision term must be rewritten with an explicit target and a consistent scalar formulation.
  3. [§4.2, §4.4, Tables 2–8] The reported numbers are obtained on the validation set (Section 4.2: "the best accuracy on the validation set was reported"), and the same validation set is used to select the patch size (Table 3), the loss balance λ (Fig. 11), the threshold t (Fig. 12), and the loss balance α (Fig. 13). Tables 2 through 8 report single runs without error bars or significance tests. The reported performance gaps over baselines may therefore partly reflect selection on the evaluation set. The authors should report results on a separately held-out test split, or use cross-validation with hyperparameters fixed independently of the test set, and provide error bars or significance measures.
  4. [§4.1, Table 1] The patch statistics in Table 1 suggest that the train/test/validation split is performed at the patch level rather than at the whole-slide level. In CAMELYON16, for example, 13,184 training patches, 3,776 test patches, and 1,883 validation patches are derived from only 270 WSIs; if patches from the same WSI appear in both training and validation, the evaluation is inflated by spatial correlation. The paper should clarify whether the split is by WSI or by patch, and if the latter is the case, the experiments should be rerun with a WSI-exclusive split.
  5. [§3.6, Eq. (27)] The pseudo-label refinement rule y_refined = β·y_cam + (1−β)·y_pred conflicts with the surrounding text. For β > 0.7, the CAM is weighted more strongly than the model prediction, yet the text states that "in regions where the model's predictions have high confidence, the model's predictions are prioritized over the CAM labels." The parameter β is also described as a confidence threshold but used as a mixing weight. This inconsistency must be resolved to make the self-correction mechanism reproducible.
minor comments (6)
  1. [§1] The introduction states that "Section 4 presents experimental results on three datasets," but Section 4.1 describes only two datasets (private glomerular lesions and CAMELYON16). Please correct the count.
  2. [§3.4.3, Eq. (13)] The notation F''[-1] in Eq. (13) is not defined; it presumably denotes the reverse of the sequence, but this should be stated explicitly.
  3. [§3.4.2, Algorithm 1] Algorithm 1 line 5 ("LearnableParameters") duplicates the learnable-weight operation already introduced in line 3 and does not match the text around Eq. (11). Please align the algorithm box with the equations.
  4. [Table 3] The rows of Table 3 are difficult to read: the columns "Patch size" and "Num_patchs" contain combined values such as "1100 22*22" and the final row "FMaMIL(100)" is not aligned with the other rows. Please reformat the table with clear, separate columns.
  5. [Figures 14 and 15] The figure captions contain typos: "FSSS" should be "FSS" or "U-Net," and "PatchCamelyon16" should be "CAMELYON16" to match the dataset name used elsewhere in the text.
  6. [§3.4.3, Eq. (14)] The use of absolute value in Eq. (14) is ambiguous because x_fft_space may be complex after the IFFT; please clarify whether the real part or the complex modulus is used.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: reported Dice/mIoU rest on independent pixel-level validation, not on the model's own pseudo-labels.

full rationale

FMaMIL is an empirical architecture paper: its central claims are experimental, and the reported segmentation metrics are evaluated against independent pixel-level annotations, not against the model's own pseudo-labels. The MIL head is trained with bag-level cross-entropy (Eq. 22) plus an instance-level term (Eq. 23) that weights each instance's cross-entropy by the attention weight; this is a self-consistency regularizer rather than a fitted target for the paper's reported numbers. The second-stage CAM-guided refinement (Eq. 27) does combine the CAM with the model's own prediction during training, but Section 4.2 states that 'All models were trained on the training set, and the best accuracy on the validation set was reported,' and Section 4.3 states that 'we invited professional doctors to provide real and detailed annotations for the validation dataset.' The final Dice/mIoU values therefore do not reduce to the pseudo-labels or attention maps by construction. The learnable frequency-domain module (Eqs. 10-12) is a trainable filter with a skip connection and does not define the target. References to prior Mamba/MIL works (e.g., [67], [15], [69]) are used as architectural inspiration or baselines, not as a load-bearing uniqueness theorem, and no cited result substitutes for the reported experiments. The CAMELYON16 patch-labeling description in Section 4.1 and the use of the validation set for hyperparameter selection are experimental-design and correctness risks, not circularity, and therefore do not affect the circularity score.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim rests on several domain assumptions: the MIL bag-to-instance relation (Eq. 5), the adequacy of frozen ResNet50 features, the validity of CAM seeds, and the safety of self-refinement. The paper also fits five hyperparameters to the validation set (patch size, lambda, t, alpha, beta), which are free parameters in the reported pipeline. No new physical entities or external mediators are introduced.

free parameters (5)
  • patch_size = 100x100
    Selected by validation performance on both datasets; Table 3 evaluates 50, 100, 150, 200, and 220, with 100 best. This is a data-fitted choice, not predetermined.
  • classification loss balance lambda = 0.2
    Tuned in [0.1, 0.9] on validation; Fig. 11 and Section 4.4.2.
  • CAM binarization threshold t = 0.5
    Tuned in [0, 1] on validation; Fig. 12 and Section 4.4.3.
  • segmentation loss balance alpha = 0.3
    Tuned in [0, 1] on validation; Fig. 13 and Section 4.4.4.
  • confidence threshold beta = > 0.7
    Chosen for confidence-weighted pseudo-label refinement in Eq. 27; no sensitivity analysis is shown.
assumptions (6)
  • domain assumption MIL assumption: a WSI bag is positive iff at least one patch instance is positive (Eq. 5).
    Section 3.1.2; the whole weakly supervised strategy rests on this.
  • domain assumption Pre-trained ResNet50 features provide sufficient representation for instance-level MIL.
    Sections 3.3 and 4.2; the frozen encoder is not trained end-to-end, so the upper bound depends on it.
  • standard math FFT/IFFT on the sequence dimension with learnable complex weights preserves invertibility and gradient flow.
    Section 3.4.2, Eqs. 10 to 12; the formulation assumes complex arithmetic and that the skip connection recovers local details.
  • domain assumption CAM from the first-stage classifier is a valid seed for segmentation pseudo-labels.
    Section 3.6; the second stage is trained entirely from first-stage CAMs.
  • ad hoc to paper Confidence-thresholded self-refinement with beta > 0.7 corrects label noise rather than amplifying it.
    Section 3.6, Eq. 27; no experiment validates that self-correction converges to ground truth rather than to the model's own bias.
  • ad hoc to paper Ground-truth annotations in the benchmarks are incomplete enough that weak supervision can outperform fully supervised training.
    Section 4.3 uses this to explain a counterintuitive result; no annotation-quality analysis is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FMaMIL: Frequency-Driven Mamba Multi-Instance Learning for Weakly Supervised Lesion Segmentation in Medical Images." pith.science (2026). https://pith.science/paper/UFUGD2MY

@misc{pith2026250607652,
  author       = {Pith},
  title        = {Pith review of: FMaMIL: Frequency-Driven Mamba Multi-Instance Learning for Weakly Supervised Lesion Segmentation in Medical Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UFUGD2MY}},
  note         = {Machine review of arXiv:2506.07652}
}
read the original abstract

Accurate lesion segmentation in histopathology images is essential for diagnostic interpretation and quantitative analysis, yet it remains challenging due to the limited availability of costly pixel-level annotations. To address this, we propose FMaMIL, a novel two-stage framework for weakly supervised lesion segmentation based solely on image-level labels. In the first stage, a lightweight Mamba-based encoder is introduced to capture long-range dependencies across image patches under the MIL paradigm. To enhance spatial sensitivity and structural awareness, we design a learnable frequency-domain encoding module that supplements spatial-domain features with spectrum-based information. CAMs generated in this stage are used to guide segmentation training. In the second stage, we refine the initial pseudo labels via a CAM-guided soft-label supervision and a self-correction mechanism, enabling robust training even under label noise. Extensive experiments on both public and private histopathology datasets demonstrate that FMaMIL outperforms state-of-the-art weakly supervised methods without relying on pixel-level annotations, validating its effectiveness and potential for digital pathology applications.

Figures

Figures reproduced from arXiv: 2506.07652 by the authors.

Figure 1
Figure 1. Comparison of three two-stage WSS frameworks for pathology images. (a) Tile classification-based method for generating pseudo masks; (b) Transformer-based MIL with attention map generation; (c) Our proposed method, which incorporates learnable frequency features and a lightweight Mamba encoder to achieve efficient and accurate mask gener￾ation with label refinement.. Image-level Supervision Classification Model Seed… view at source ↗
Figure 2
Figure 2. illustrates the common diffuse characteristics of lesions. Hangbei Cheng et al.: Preprint submitted to Elsevier Page 17 of 17 [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗
Figure 3
Figure 3. The overall framework of our proposed FMaMIL primarily consists of three stages. Pretrained Model Sentence Reordering Cls token 1 10 19 … 73 2 11 … 80 9 18 … 63 72 Cls token 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 19 20 21 22 23 24 25 26 27 12 13 14 15 16 17 4 5 6 7 8 75 76 77 77 Row Scan Col Scan 1 2 3 4 n-1 n cls 1 2 3 4 ..… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Sentence Reordering Structure. Input Embedding Frequency-domian Feature W Learning Frequency FFT Feature IFFT B× d× L B× d×H×W B× d×H×W Learnable Frequency-domain Encoding + [PITH_FULL_IMAGE:figures/full_fig_p019_4.png]
Figure 7
Figure 7. Figure 7: Instance-level to Bag-level Classification Head Block Structure. Hangbei Cheng et al.: Preprint submitted to Elsevier Page 18 of 17 [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 11
Figure 11. Figure 11: classification Loss of balance 𝜆 (a) GLO (b) Camelyon16 [PITH_FULL_IMAGE:figures/full_fig_p020_11.png]
Figure 9
Figure 9. Figure 9: Dataset samples (a) (b) (c) (d) (e) (f) [PITH_FULL_IMAGE:figures/full_fig_p020_9.png]
Figure 10
Figure 10. Figure 10: Visualization results of lesion areas with different patch sizes. (a) Original image; (b) Patch size = 50; (c) Patch size = 100 (d) Patch size = 150; (e) Patch size = 200; (f) Patch size = 220; 𝜆 𝜆 (a) Dice (b) mIoU [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 14
Figure 14. Figure 14: Visualization comparison of lesion segmentation of Glo (a) Image (b) Label (c) Ours (d) FSSS (e) Auxsegnet (f) Group-WSSS (g)SEAM (h) SA-MIL (i) TransMIL (j) MambaMIL (k) ConvNext [PITH_FULL_IMAGE:figures/full_fig_p021_14.png]
Figure 15
Figure 15. Figure 15: Visualization comparison of lesion segmentation of PatchCamelyon16 Hangbei Cheng et al.: Preprint submitted to Elsevier Page 20 of 17 [PITH_FULL_IMAGE:figures/full_fig_p021_15.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

85 extracted references · 61 canonical work pages

  1. [1]

    Single-stage semantic segmentation from image labels, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Araslanov, N., Roth, S., 2020. Single-stage semantic segmentation from image labels, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 4253–4262

    2020

  2. [2]

    Weakly and semi supervised detection in medical imaging via deep dual branch net

    Bakalo, R., Goldberger, J., Ben-Ari, R., 2021. Weakly and semi supervised detection in medical imaging via deep dual branch net. Neurocomputing 421, 15–25

    2021

  3. [3]

    What’s the point: Semantic segmentation with point supervision, in: Euro- pean conference on computer vision, Springer

    Bearman, A., Russakovsky, O., Ferrari, V., Fei-Fei, L., 2016. What’s the point: Semantic segmentation with point supervision, in: Euro- pean conference on computer vision, Springer. pp. 549–565

    2016

  4. [4]

    Clinical-grade computational pathology using weakly supervised deep learning on whole slide images

    Campanella, G., Hanna, M.G., Geneslaw, L., Miraflor, A., Werneck KraussSilva,V.,Busam,K.J.,Brogi,E.,Reuter,V.E.,Klimstra,D.S., Fuchs, T.J., 2019. Clinical-grade computational pathology using weakly supervised deep learning on whole slide images. Nature medicine 25, 1301–1309

    2019

  5. [5]

    Multi-branch spectral channel attention network for breast cancer histopathology image classification

    Cao, L., Pan, K., Ren, Y., Lu, R., Zhang, J., 2024. Multi-branch spectral channel attention network for breast cancer histopathology image classification. Electronics 13, 459

    2024

  6. [6]

    Chang, A., Zeng, J., Huang, R., Ni, D., 2024. Em-net: Efficient channel and frequency learning with mamba for 3d medical image segmentation, in: International Conference on Medical Image Com- puting and Computer-Assisted Intervention, Springer. pp. 266–275

    work page 2024

  7. [7]

    Encoder-decoder with atrous separable convolution for semantic im- age segmentation, in: Proceedings of the European conference on computer vision (ECCV), pp

    Chen, L.C., Zhu, Y., Papandreou, G., Schroff, F., Adam, H., 2018. Encoder-decoder with atrous separable convolution for semantic im- age segmentation, in: Proceedings of the European conference on computer vision (ECCV), pp. 801–818

    work page 2018

  8. [8]

    Multiple instance learning with bag dissimilarities

    Cheplygina, V., Tax, D.M., Loog, M., 2015. Multiple instance learning with bag dissimilarities. Pattern recognition 48, 264–275

    work page 2015

Show all 85 references
  1. [9]

    Boxsup: Exploiting bounding boxes to supervise convolutional networks for semantic segmentation, in: ProceedingsoftheIEEEinternationalconferenceoncomputervision, pp

    Dai, J., He, K., Sun, J., 2015. Boxsup: Exploiting bounding boxes to supervise convolutional networks for semantic segmentation, in: ProceedingsoftheIEEEinternationalconferenceoncomputervision, pp. 1635–1643

    2015

  2. [10]

    Solving the multiple instance problem with axis-parallel rectangles

    Dietterich, T.G., Lathrop, R.H., Lozano-Pérez, T., 1997. Solving the multiple instance problem with axis-parallel rectangles. Artificial intelligence 89, 31–71

    1997

  3. [11]

    Slf-unet: Improved unet for brain mri segmentation by combining spatial and low-frequency domain features, in: Computer Graphics International Conference, Springer

    Ding, H., Lu, J., Cai, J., Zhang, Y., Shang, Y., 2023. Slf-unet: Improved unet for brain mri segmentation by combining spatial and low-frequency domain features, in: Computer Graphics International Conference, Springer. pp. 415–426

    2023

  4. [12]

    Dong, N., Kampffmeyer, M., Liang, X., Wang, Z., Dai, W., Xing, E., 2018. Reinforced auto-zoom net: towards accurate and fast breastcancersegmentationinwhole-slideimages,in:DeepLearning in Medical Image Analysis and Multimodal Learning for Clinical DecisionSupport:4thInternatio...

    2018

  5. [13]

    Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer

    Ehteshami Bejnordi, B., Veta, M., Johannes van Diest, P., van Ginneken, B., Karssemeijer, N., Litjens, G., van der Laak, J.A.W.M., , the CAMELYON16 Consortium, 2017. Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast c...

    2017

  6. [14]

    Fang,H.,Huang,S.,Tang,W.,Huangfu,L.,Liu,B.,2024a. Sam-mil: A spatial contextual aware multiple instance learning approach for whole slide image classification, in: Proceedings of the 32nd ACM International Conference on Multimedia, pp. 6083–6092

  7. [15]

    Mammil: Multiple instance learning for whole slide images with state space models, in: 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), IEEE

    Fang,Z.,Wang,Y.,Zhang,Y.,Wang,Z.,Zhang,J.,Ji,X.,Zhang,Y., 2024b. Mammil: Multiple instance learning for whole slide images with state space models, in: 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), IEEE. pp. 3200–3205

    2024

  8. [16]

    Fiba: Frequency-injectionbasedbackdoorattackinmedicalimageanalysis, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Feng, Y., Ma, B., Zhang, J., Zhao, S., Xia, Y., Tao, D., 2022. Fiba: Frequency-injectionbasedbackdoorattackinmedicalimageanalysis, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 20876–20885

    2022

  9. [17]

    Mamba: Linear-time sequence modeling with selective state spaces

    Gu, A., Dao, T., 2023. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752

    2023 arXiv

  10. [18]

    Hippo:Recurrent memory with optimal polynomial projections

    Gu,A.,Dao,T.,Ermon,S.,Rudra,A.,Ré,C.,2020. Hippo:Recurrent memory with optimal polynomial projections. Advances in neural information processing systems 33, 1474–1487

    2020

  11. [19]

    Asurveyonvisiontransformer

    Han, K., Wang, Y., Chen, H., Chen, X., Guo, J., Liu, Z., Tang, Y., Xiao,A.,Xu,C.,Xu,Y.,etal.,2022. Asurveyonvisiontransformer. IEEE transactions on pattern analysis and machine intelligence 45, 87–110

    2022

  12. [20]

    T-mamba: Frequency-enhanced gated long-range dependency for tooth 3d cbct segmentation

    Hao, J., He, L., Hung, K.F., 2024. T-mamba: Frequency-enhanced gated long-range dependency for tooth 3d cbct segmentation. arXiv preprint arXiv:2404.01065

    2024 arXiv

  13. [21]

    Deep residual learning for image recognition, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp

    He, K., Zhang, X., Ren, S., Sun, J., 2016. Deep residual learning for image recognition, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770–778

    2016

  14. [22]

    Hou, L., Samaras, D., Kurc, T.M., Gao, Y., Davis, J.E., Saltz, J.H.,

  15. [23]

    State-space models are accurate and efficient neural operators for dynamical systems

    Hu, Z., Daryakenari, N.A., Shen, Q., Kawaguchi, K., Karniadakis, G.E., 2024. State-space models are accurate and efficient neural operators for dynamical systems. arXiv preprint arXiv:2409.03231

    2024 arXiv

  16. [24]

    Attention-based deep multiple instance learning, in: International conference on machine learning, PMLR

    Ilse, M., Tomczak, J., Welling, M., 2018. Attention-based deep multiple instance learning, in: International conference on machine learning, PMLR. pp. 2127–2136

    2018

  17. [25]

    Online attentionaccumulationforweaklysupervisedsemanticsegmentation

    Jiang, P.T., Han, L.H., Hou, Q., Cheng, M.M., Wei, Y., 2021. Online attentionaccumulationforweaklysupervisedsemanticsegmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence 44, 7062–7077

    2021

  18. [26]

    Jiang, P.T., Hou, Q., Cao, Y., Cheng, M.M., Wei, Y., Xiong, H.K.,

  19. [27]

    Jiang, P.T., Yang, Y., Hou, Q., Wei, Y., 2022. L2g: A simple local-to-global knowledge transfer framework for weakly supervised semanticsegmentation,in:ProceedingsoftheIEEE/CVFconference on computer vision and pattern recognition, pp. 16886–16896

    2022

  20. [28]

    Weakly-supervised learning for lung carcinoma classification using deep learning

    Kanavati, F., Toyokawa, G., Momosaki, S., Rambeau, M., Kozuma, Y., Shoji, F., Yamazaki, K., Takeo, S., Iizuka, O., Tsuneki, M., 2020. Weakly-supervised learning for lung carcinoma classification using deep learning. Scientific reports 10, 9297

    2020

  21. [29]

    Khan, S., Naseer, M., Hayat, M., Zamir, S.W., Khan, F.S., Shah, M.,

  22. [30]

    Lerousseau, M., Vakalopoulou, M., Classe, M., Adam, J., Battistella, E., Carré, A., Estienne, T., Henry, T., Deutsch, E., Paragios, N., 2020.Weaklysupervisedmultipleinstancelearninghistopathological tumor segmentation, in: Medical Image Computing and Computer AssistedIntervent...

    2020

  23. [31]

    Videomamba:Statespacemodelforefficientvideounderstanding,in: European Conference on Computer Vision, Springer

    Li, K., Li, X., Wang, Y., He, Y., Wang, Y., Wang, L., Qiao, Y., 2025. Videomamba:Statespacemodelforefficientvideounderstanding,in: European Conference on Computer Vision, Springer. pp. 237–255

    2025

  24. [32]

    Weakly supervised histopathology image segmentation with self-attention

    Li, K., Qian, Z., Han, Y., Eric, I., Chang, C., Wei, B., Lai, M., Liao, J., Fan, Y., Xu, Y., 2023. Weakly supervised histopathology image segmentation with self-attention. Medical Image Analysis 86, 102791

    2023

  25. [33]

    Group-wise semantic mining for weakly supervised semantic segmentation, in: Proceedings of the AAAI conference on artificial intelligence, pp

    Li, X., Zhou, T., Li, J., Zhou, Y., Zhang, Z., 2021. Group-wise semantic mining for weakly supervised semantic segmentation, in: Proceedings of the AAAI conference on artificial intelligence, pp. 1984–1992

    2021

  26. [34]

    Modeling annotator prefer- enceandstochasticannotationerrorformedicalimagesegmentation

    Liao, Z., Hu, S., Xie, Y., Xia, Y., 2024. Modeling annotator prefer- enceandstochasticannotationerrorformedicalimagesegmentation. Medical Image Analysis 92, 103028

    2024

  27. [35]

    Scan- net:Afastanddensescanningframeworkformetastasticbreastcancer detection from whole-slide image, in: 2018 IEEE winter conference on applications of computer vision (WACV), IEEE

    Lin,H.,Chen,H.,Dou,Q.,Wang,L.,Qin,J.,Heng,P.A.,2018. Scan- net:Afastanddensescanningframeworkformetastasticbreastcancer detection from whole-slide image, in: 2018 IEEE winter conference on applications of computer vision (WACV), IEEE. pp. 539–546. Hangbei Cheng et al.:Preprint...

    2018

  28. [36]

    Litjens, G., Kooi, T., Bejnordi, B.E., Setio, A.A.A., Ciompi, F., Ghafoorian,M.,VanDerLaak,J.A.,VanGinneken,B.,Sánchez,C.I.,

  29. [37]

    Liu, Q., Chen, C., Qin, J., Dou, Q., Heng, P.A., 2021a. Feddg: Federated domain generalization on medical image segmentation via episodic learning in continuous frequency space, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recogni- tion, pp. 1013–1023

  30. [38]

    A review of deep- learning-based medical image segmentation methods

    Liu, X., Song, L., Liu, S., Zhang, Y., 2021b. A review of deep- learning-based medical image segmentation methods. Sustainability 13, 1224

  31. [39]

    Liu, Y., Tian, Y., Zhao, Y., Yu, H., Xie, L., Wang, Y., Ye, Q., Liu, Y.,

  32. [40]

    Liu, Z., Mao, H., Wu, C.Y., Feichtenhofer, C., Darrell, T., Xie, S.,

  33. [41]

    Decoupled weight decay regularization

    Loshchilov, I., 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101

    2017 arXiv

  34. [42]

    Data-efficient and weakly supervised com- putational pathology on whole-slide images

    Lu, M.Y., Williamson, D.F., Chen, T.Y., Chen, R.J., Barbieri, M., Mahmood, F., 2021. Data-efficient and weakly supervised com- putational pathology on whole-slide images. Nature biomedical engineering 5, 555–570

    2021

  35. [43]

    A framework for multiple- instancelearning

    Maron, O., Lozano-Pérez, T., 1997. A framework for multiple- instancelearning. Advancesinneuralinformationprocessingsystems 10

    1997

  36. [44]

    Qu, H., Ning, L., An, R., Fan, W., Derr, T., Liu, H., Xu, X., Li, Q.,

  37. [45]

    Bi-directional weakly supervisedknowledgedistillationforwholeslideimageclassification

    Qu, L., Wang, M., Song, Z., et al., 2022. Bi-directional weakly supervisedknowledgedistillationforwholeslideimageclassification. Advances in Neural Information Processing Systems 35, 15368– 15381

    2022

  38. [46]

    11976– 11986

    A convnet for the 2020s, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 11976– 11986

  39. [47]

    Ronneberger, O., Fischer, P., Brox, T., 2015. U-net: Convolutional networks for biomedical image segmentation, in: Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, pro- ceedings, part III...

    2015

  40. [48]

    Learning multi-axis representation in frequency domain for medical image segmentation

    Ruan, J., Gao, J., Xie, M., Xiang, S., 2025. Learning multi-axis representation in frequency domain for medical image segmentation. Machine Learning 114, 10

    2025

  41. [49]

    Transmil:Transformerbasedcorrelatedmultipleinstancelearningfor whole slide image classification

    Shao, Z., Bian, H., Chen, Y., Wang, Y., Zhang, J., Ji, X., et al., 2021. Transmil:Transformerbasedcorrelatedmultipleinstancelearningfor whole slide image classification. Advances in neural information processing systems 34, 2136–2147

    2021

  42. [50]

    Attention is all you need

    Vaswani, A., 2017. Attention is all you need. Advances in Neural Information Processing Systems

    2017

  43. [51]

    arXiv preprint arXiv:2408.01129

    A survey of mamba. arXiv preprint arXiv:2408.01129

    arXiv

  44. [52]

    Wang, L., Lu, H., Wang, Y., Feng, M., Wang, D., Yin, B., Ruan, X.,

  45. [53]

    IEEEreviews in biomedical engineering 10, 213–234

    Quellec,G.,Cazuguel,G.,Cochener,B.,Lamard,M.,2017.Multiple- instancelearningformedicalimageandvideoanalysis. IEEEreviews in biomedical engineering 10, 213–234

    2017

  46. [54]

    The American journal of pathology 189, 1686–1698

    Wang,S.,Yang,D.M.,Rong,R.,Zhan,X.,Xiao,G.,2019a.Pathology image analysis using segmentation deep learning algorithms. The American journal of pathology 189, 1686–1698

  47. [55]

    Weakly supervised deep learning for whole slide lung cancer image analysis

    Wang,X.,Chen,H.,Gan,C.,Lin,H.,Dou,Q.,Tsougenis,E.,Huang, Q., Cai, M., Heng, P.A., 2019b. Weakly supervised deep learning for whole slide lung cancer image analysis. IEEE transactions on cybernetics 50, 3950–3962

  48. [56]

    Ud-mil: uncertainty-driven deep multiple instance learning for oct image classification

    Wang, X., Tang, F., Chen, H., Luo, L., Tang, Z., Ran, A.R., Cheung, C.Y., Heng, P.A., 2020. Ud-mil: uncertainty-driven deep multiple instance learning for oct image classification. IEEE journal of biomedical and health informatics 24, 3431–3442

    2020

  49. [57]

    State space model for new-generation network alternative to transformers: A survey

    Wang, X., Wang, S., Ding, Y., Li, Y., Wu, W., Rong, Y., Kong, W., Huang, J., Li, S., Yang, H., et al., 2024. State space model for new-generation network alternative to transformers: A survey. arXiv preprint arXiv:2404.09516

    2024 arXiv

  50. [58]

    Anempiricalstudyofmamba-basedlanguagemodels

    Waleffe, R., Byeon, W., Riach, D., Norick, B., Korthikanti, V., Dao, T., Gu, A., Hatamizadeh, A., Singh, S., Narayanan, D., et al., 2024. Anempiricalstudyofmamba-basedlanguagemodels. arXivpreprint arXiv:2406.07887

    2024 arXiv

  51. [59]

    Wei, Y., Feng, J., Liang, X., Cheng, M.M., Zhao, Y., Yan, S., 2017. Objectregionminingwithadversarialerasing:Asimpleclassification to semantic segmentation approach, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1568– 1576

    2017

  52. [60]

    Learningtodetectsalientobjectswithimage-levelsupervision, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 136–145

  53. [61]

    Annotation-efficientdeeplearning for automatic medical image segmentation

    Wang, S., Li, C., Wang, R., Liu, Z., Wang, M., Tan, H., Wu, Y., Liu, X.,Sun,H.,Yang,R.,etal.,2021. Annotation-efficientdeeplearning for automatic medical image segmentation. Nature communications 12, 5915

    2021

  54. [62]

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

    Xu, H., Usuyama, N., Bagga, J., Zhang, S., Rao, R., Naumann, T., Wong, C., Gero, Z., González, J., Gu, Y., et al., 2024. A whole-slide foundation model for digital pathology from real-world data. Nature , 1–8

    2024

  55. [63]

    Learning in the frequency domain, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp

    Xu, K., Qin, M., Sun, F., Wang, Y., Chen, Y.K., Ren, F., 2020. Learning in the frequency domain, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 1740– 1749

    2020

  56. [64]

    Xu,L.,Ouyang,W.,Bennamoun,M.,Boussaid,F.,Sohel,F.,Xu,D.,

  57. [65]

    Large scale tissue histopathology image classification, segmentation, and visualization via deep convolutional activation features

    Xu, Y., Jia, Z., Wang, L.B., Ai, Y., Zhang, F., Lai, M., Chang, E.I.C., 2017. Large scale tissue histopathology image classification, segmentation, and visualization via deep convolutional activation features. BMC bioinformatics 18, 1–17

    2017

  58. [66]

    Revisiting multiple instance neural networks

    Wang, X., Yan, Y., Tang, P., Bai, X., Liu, W., 2018. Revisiting multiple instance neural networks. Pattern recognition 74, 15–24

    2018

  59. [67]

    Yang, S., Wang, Y., Chen, H., 2024. Mambamil: Enhancing long se- quence modeling with sequence reordering in computational pathol- ogy, in: International Conference on Medical Image Computing and Computer-Assisted Intervention, Springer. pp. 296–306

    2024

  60. [68]

    Wu, Y., Ye, X., Yang, K., Li, J., Li, X., 2024. Dupl: Dual student with trustworthy progressive learning for robust weakly supervised semanticsegmentation,in:ProceedingsoftheIEEE/CVFConference on Computer Vision and Pattern Recognition, pp. 3534–3543

    2024

  61. [69]

    Learningfrom massivenoisylabeleddataforimageclassification,in:Proceedingsof the IEEE conference on computer vision and pattern recognition, pp

    Xiao,T.,Xia,T.,Yang,Y.,Huang,C.,Wang,X.,2015. Learningfrom massivenoisylabeleddataforimageclassification,in:Proceedingsof the IEEE conference on computer vision and pattern recognition, pp. 2691–2699

    2015

  62. [70]

    Attention-challenging multiple instance learning for whole slide im- age classification, in: European Conference on Computer Vision, Springer

    Zhang, Y., Li, H., Sun, Y., Zheng, S., Zhu, C., Yang, L., 2025. Attention-challenging multiple instance learning for whole slide im- age classification, in: European Conference on Computer Vision, Springer. pp. 125–143

    2025

  63. [71]

    Mamba2mil: State space duality based multiple instance learning for computational pathology

    Zhang, Y., Zhang, X., Wang, J., Yang, Y., Peng, T., Tong, C., 2024. Mamba2mil: State space duality based multiple instance learning for computational pathology. arXiv preprint arXiv:2408.15032

    2024 arXiv

  64. [72]

    Zheng, Y., Sharma, H., Betke, M., Beane, J., Kolachalama, V.B.,

  65. [73]

    Learning deep features for discriminative localization, in: Proceed- ings of the IEEE conference on computer vision and pattern recogni- tion, pp

    Zhou, B., Khosla, A., Lapedriza, A., Oliva, A., Torralba, A., 2016. Learning deep features for discriminative localization, in: Proceed- ings of the IEEE conference on computer vision and pattern recogni- tion, pp. 2921–2929. Hangbei Cheng et al.:Preprint submitted to Elsevier...

    2016

  66. [74]

    Spatial- frequency dual progressive attention network for medical image seg- mentation

    Zhou, Z., He, A., Wu, Y., Yao, R., Xie, X., Li, T., 2024. Spatial- frequency dual progressive attention network for medical image seg- mentation. arXiv preprint arXiv:2406.07952

    2024 arXiv

  67. [75]

    Weakly supervisedhistopathology cancerimage segmentationand classifica- tion

    Xu, Y., Zhu, J.Y., Eric, I., Chang, C., Lai, M., Tu, Z., 2014. Weakly supervisedhistopathology cancerimage segmentationand classifica- tion. Medical image analysis 18, 591–604

    2014

  68. [76]

    Vision mamba: Efficient visual representation learning with bidirec- tional state space model

    Zhu, L., Liao, B., Zhang, Q., Wang, X., Liu, W., Wang, X., 2024b. Vision mamba: Efficient visual representation learning with bidirec- tional state space model. arXiv preprint arXiv:2401.09417 . First Stage Second Stage ⋯ Image-level ⋯ Spatial domain Feature extraction Tiles C...

    arXiv

  69. [77]

    Whole slide images based cancer survival prediction using attention guided deep multiple instance learning networks

    Yao, J., Zhu, X., Jonnagaddala, J., Hawkins, N., Huang, J., 2020. Whole slide images based cancer survival prediction using attention guided deep multiple instance learning networks. Medical image analysis 65, 101789

    2020

  70. [78]

    Medmamba: Vision mamba for medical image classification

    Yue, Y., Li, Z., 2024. Medmamba: Vision mamba for medical image classification. arXiv preprint arXiv:2403.03849

    2024 arXiv

  71. [82]

    bioRxiv , 2024–08

    Fouriermil: Fourier filtering-based multiple instance learning for whole slide image analysis. bioRxiv , 2024–08

    2024

  72. [2016]

    2424–2433

    Patch-based convolutional neural network for whole slide tissue image classification, in: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2424–2433

  73. [2017]

    Medical image analysis 42, 60–88

    A survey on deep learning in medical image analysis. Medical image analysis 42, 60–88

  74. [2019]

    2070–2079

    Integral object mining via online attention accumulation, in: Proceedings of the IEEE/CVF international conference on computer vision, pp. 2070–2079

  75. [2021]

    6984–6993

    Leveraging auxiliary tasks with affinity learning for weakly supervisedsemanticsegmentation,in:ProceedingsoftheIEEE/CVF international conference on computer vision, pp. 6984–6993

  76. [2022]

    ACM computing surveys (CSUR) 54, 1–41

    Transformers in vision: A survey. ACM computing surveys (CSUR) 54, 1–41

  77. [2024]

    CoRR abs/2401.10166

    Vmamba: Visual state space model. CoRR abs/2401.10166. URL:https://doi.org/10.48550/arXiv.2401.10166

    doi arXiv

Pith tools

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