Pith. sign in

REVIEW 5 major objections 5 minor 51 references

FECT: Classification of Breast Cancer Pathological Images Based on Fusion Features

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read FECT, a model that fuses cell, tissue, and edge features, reports a new best weighted F1 of 65.8% on the seven-class BRACS breast tissue classification benchmark.

desk verdict A plausible, modest benchmark gain on BRACS, but the edge-feature branch relies on the authors' unreleased segmentation masks and the SOTA claim is not yet isolated from that extra supervision. read the letter →

arxiv 2501.10128 v1 pith:YQHNDAV6 submitted 2025-01-17 eess.IV cs.CV

classification eess.IVcs.CV
keywords BreastcancerclassificationFeaturefusionTransformerAttention-basedaggregatorMultimodalBRACSdatasetPathologicalimageEdgefeatures
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

Breast tissue diagnosis requires looking at different scales: individual cells, overall tissue architecture, and the boundary where a lesion meets surrounding tissue. This paper argues that existing classifiers rely on only one of these scales, and that a model which extracts and fuses all three can do better. The authors build FECT, which segments cells with HoverNet, encodes tissue structure with ResMTUNet, samples patches along tissue edges and builds a cell-adjacency graph at those edges, then weights and concatenates the three feature sets into an SVM classifier. On the seven-category BRACS benchmark, FECT reports a weighted F1 score of 65.8%—an average of per-category F1 scores weighted by class size—and an accuracy of 66.37%, ahead of the compared methods, with the clearest gains in normal tissue, ductal carcinoma in situ, and invasive carcinoma. If the result holds, the takeaway is that boundary-level morphology—the feature pathologists use to separate invasive from in situ disease—deserves a place in automated classifiers, not just cell and tissue features.

What carries the argument

The load-bearing mechanism is the three-branch feature extraction and fusion pipeline. Each branch is designed to capture one scale a pathologist attends to: cells (nucleus segmentation plus attention-based aggregation), tissue (ResMTUNet global architecture), and edges (contour-sampled patches encoded by a vision transformer, aggregated with Nyström attention, and linked by a KNN adjacency graph of cells near the boundary). The fused representation is a weighted concatenation $X_f = [\alpha X_C, \beta X_T, \gamma X_E]$ fed to an SVM. The ablation design is what carries the argument: comparing single branches against their combinations shows that the full fusion outperforms every subset, and that the edge branch contributes most when the other two are present.

What would settle it

Replace the semi-automated seven-category masks with automatically predicted masks and rerun the edge branch; if the weighted F1 advantage over ScoreNet disappears, the claimed benefit of the fusion architecture is actually the benefit of the extra mask supervision.

Watch

Extended reading notes

Core claim

The paper's central claim is that FECT—Fused features of Edges, Cells, and Tissues—sets a new best result on the BRACS Tissue Regions of Interest task by combining three complementary views of each image. The cell branch uses HoverNet to segment nuclei and an attention aggregator to combine per-cell embeddings. The tissue branch uses ResMTUNet to encode overall tissue architecture. The edge branch samples 64×64 patches along the boundaries of the tissue mask, encodes them with a vision transformer pretrained on ImageNet, aggregates them with Nyström attention, and constructs a KNN graph of cells at those edges to capture the tumor microenvironment. The three feature vectors are weighted, concatenated, and classified by an SVM. On the BRACS test set, FECT achieves a weighted F1 of 65.8% and accuracy of 66.37%, surpassing the compared methods, including HACT-Net (61.5% weighted F1) and ScoreNet (64.4% weighted F1), and the ablation study shows that edge features, although weak on their own, improve classification when added to cell and tissue features, especially for ductal carcinoma in situ and invasive carcinoma.

Load-bearing premise

The edge-feature branch is built on precise seven-category segmentation masks that were generated semi-automatically in prior work, and the paper never tests whether the performance gain comes from having those masks rather than from the fusion design itself.

Editorial extensions

If this is right

  • If FECT's reported numbers replicate, the 65.8% weighted F1 is the new reference point for seven-class BRACS tissue classification, ahead of the 64.4% of ScoreNet and 61.5% of HACT-Net.
  • The ablations show the fused model beats every single-feature and every two-feature combination, so the paper's design argument is that the three scales are genuinely complementary.
  • Edge features, while the weakest branch on their own, are what push ductal carcinoma in situ and invasive carcinoma apart in the fused model; any future method targeting that pair should include boundary cues.
  • The SVM-with-weighted-concatenation configuration outperforms the other classifiers tested on the same fused features, making the fusion-and-classifier combination part of the claimed result.

Reading between the lines

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

  • A direct test of the edge branch's portability is to replace the semi-automated seven-category masks with automatically predicted or coarser masks and measure how much of the FECT gain remains.
  • The margin over the closest compared method is about 1.4 weighted-F1 points, so a multi-seed replication with confidence intervals would be a natural next step before drawing strong clinical conclusions.
  • Because the invasive-versus-in-situ distinction is the clinical question the edge branch targets, the same edge-graph idea could transfer to other tumor types where invasion status matters.
  • A pathologist study could check whether the edge regions FECT attends to correspond to the areas a pathologist would confirm with myoepithelial markers such as P63.
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 / 5 minor

Summary. FECT proposes a multimodal architecture for seven-class breast tissue classification on BRACS ROIs. It extracts cell features via HoverNet segmentation and an attention-based aggregator; tissue features via ResMTUNet; and edge features via uniform sampling along contours of tissue masks, Nyström-attention aggregation of edge patches, and a KNN-based graph over cells at edges. The three feature vectors are weighted-concatenated and classified with an SVM. On BRACS, the paper reports a weighted F1 of 65.8% (Acc 66.37%), claiming a new state-of-the-art, and ablation experiments showing all three feature types contribute.

Significance. The interpretability motivation, matching pathologists' use of edge and stromal cues, is attractive, and a systematic ablation across feature subsets (Table 3) plus per-class error bars (Table 1) are useful assets. If the result holds, a 65.8% weighted F1 on the BRACS seven-class task would be a meaningful benchmark. However, the central SOTA claim is confounded by the use of extra mask supervision unavailable to baselines, and several methodological details are missing, so the current form does not yet establish the architectural advantage.

major comments (5)
  1. [Section 3.4 and Section 3.6, Table 3] The edge branch is built on the 'seven-category segmentation masks' from the authors' own prior work [19], which are semi-automatically generated on BRACS. None of the baselines in Table 1 have access to these masks. Table 3 ablates edge features as a whole but does not separate the value of the mask-supervision from the value of the proposed edge-patch/KNN-graph aggregation. Since the Cell+Tissue baseline is 62.48% F1 and adding edge features gives 65.75%, the +3.27 pt could in principle be obtained by feeding the masks (or simple mask-derived features) to a baseline. Without a control that injects the same mask information into a baseline model, or a FECT variant that uses the masks without the edge graph/aggregator, the claim that the proposed architecture is superior is not supported.
  2. [Section 4.2, Table 1] The manuscript does not state whether the baseline numbers were obtained by re-running the methods under the same experimental protocol (same BRACS train/test split, same 512×512 resizing, same preprocessing and augmentation) or are copied from the original papers. If they are copied, differences in evaluation setup could change the margins; if they are re-run, details of the reimplementation and hyperparameters should be given. The authors should either re-run all baselines under identical conditions and report them, or explicitly cite the source for each column and justify comparability.
  3. [Section 3.5 and Section 4.3] The fusion weights α, β, γ are selected by grid search, but the paper never states whether the search was performed on a held-out validation split or on the test set, and it does not report the selected values or the edge weight γ. If the grid search was done on the test set, the reported 65.75% F1 is optimistically biased. The exact validation protocol and the chosen weights must be disclosed.
  4. [Section 4.1, Tables 2–4] 'Overall classification accuracy (Acc)' is defined as 'the unweighted average of recall for each category', i.e., balanced accuracy, not the standard fraction of correctly classified samples. This mismatch is carried through Tables 2–4 and makes the 'Acc' column non-comparable to accuracy values in prior work. The authors should either use standard overall accuracy or rename the metric, and should state which version is used in each table.
  5. [Section 3.4] The edge extraction procedure is not reproducible as described. The number and density of contour sampling points, the parameter K in the KNN graph, and the way the edge graph is pooled into a feature vector (no GNN layer or readout is specified) are all missing. Moreover, the segmentation masks from [19] are not included in the Data Availability statement, so a third party cannot build the edge branch without re-deriving the masks. At minimum, the missing hyperparameters should be provided and the masks (or a detailed protocol for regenerating them) should be released.
minor comments (5)
  1. [Abstract] The abstract contains the grammatical error 'we proposes'; it should be 'we propose'.
  2. [Section 3.6] The phrase 'seven-category segmentation masks ... as proposed in [19]' is opaque; the cited paper's title concerns invasive carcinoma segmentation, so the reader cannot tell whether [19] actually provides multi-class masks or just the segmentation method. Clarify the provenance and contents of the masks.
  3. [Section 4.2] The statement that FECT shows 'exceptional performance' on ADH should be tempered because Table 1 shows ScoreNet's ADH F1 (46.7) is higher than FECT's (45.2).
  4. [Section 4.1] When describing the optimizer, consider specifying whether weight decay is used and whether any warm-up schedule accompanies the step-decay learning rate.
  5. [Table 1] The legend says the best results are bold and second-best underlined, but in the extracted text the underlining is not visible; please ensure the table rendering is unambiguous.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: FECT's SOTA claim is measured against external BRACS labels, and no equation reduces the result to its own inputs.

full rationale

The central claim (Section 4.2, Table 1) is a weighted F1-score of 65.8% on the BRACS test set, whose seven class labels come from the external BRACS dataset [20], not from FECT's own features or fitted parameters. The edge branch does use segmentation masks from the authors' prior work [19], but Section 3.6 describes those masks as produced by a semi-automated annotation process over BRACS tissue, and they are not the classification target. No equation in the paper defines the predicted label in terms of the masks, and no fitted parameter is renamed as a prediction: the fusion weights are tuned on validation behavior via grid search and then evaluated on the held-out test set, which is standard practice. The self-citations to [19] and ResMTUNet are real dependencies for reproducibility, but they are not load-bearing in the sense of forcing the benchmark outcome by construction; the comparison to prior methods is an external, empirical test. The lack of released masks and the absence of an ablation isolating mask availability from the edge-feature gain are legitimate correctness and reproducibility concerns, but they are not circularity under the specified definitions.

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

The central claim depends on the quality and availability of the segmentation masks from the authors' prior work [19], on the transferability of pretrained components, and on several hyperparameters that are not fully disclosed. No new physical entities are introduced.

free parameters (6)
  • alpha (cell feature fusion weight) = ~0.6
    Grid search over weighted fusion in Section 4.3; optimal region around cell weight 0.6 and tissue weight 0.4.
  • beta (tissue feature fusion weight) = ~0.4
    Grid search in Section 4.3.
  • gamma (edge feature fusion weight) = not reported
    Section 3.5 defines X_f = [alpha X_C, beta X_T, gamma X_E], but no value or search details for gamma are given.
  • K (number of neighbors in KNN adjacency graph) = not reported
    Section 3.4 constructs a KNN graph of cells at edges but does not state K.
  • edge sampling density = not reported
    Section 3.4 uniformly samples contour points to reduce compute; sampling rate is not specified.
  • edge patch size = 64x64
    Chosen by hand in Section 3.4; not justified against other sizes.
assumptions (5)
  • domain assumption The segmentation masks from [19] accurately delineate epithelial tissue in BRACS ROIs.
    Section 3.6 introduces masks as supervision; Section 3.4 uses them to sample edge patches and construct the edge graph. These masks come from the authors' own prior work and are not available to baseline methods.
  • domain assumption HoverNet provides reliable cell nuclei segmentation on BRACS H&E images.
    Section 3.2 uses HoverNet masks to define cell regions and train the cell feature extractor.
  • domain assumption ImageNet-pretrained ViT features are informative for 64x64 pathology edge patches.
    Section 3.4 trains an edge patch aggregator on top of ImageNet-pretrained ViT; no fine-tuning on pathology-specific data is described.
  • domain assumption KNN cell adjacency at tissue edges captures microenvironment patterns that separate IC from DCIS.
    Section 3.4 and highlight 2 claim this; the ablation in Table 3 shows edge features help only in combination, not alone.
  • domain assumption BRACS ground-truth labels are correct and consistent with the segmentation masks.
    Section 3.6 trusts the public dataset annotations; label noise would directly affect the reported F1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FECT: Classification of Breast Cancer Pathological Images Based on Fusion Features." pith.science (2026). https://pith.science/paper/YQHNDAV6

@misc{pith2026250110128,
  author       = {Pith},
  title        = {Pith review of: FECT: Classification of Breast Cancer Pathological Images Based on Fusion Features},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YQHNDAV6}},
  note         = {Machine review of arXiv:2501.10128}
}
read the original abstract

Breast cancer is one of the most common cancers among women globally, with early diagnosis and precise classification being crucial. With the advancement of deep learning and computer vision, the automatic classification of breast tissue pathological images has emerged as a research focus. Existing methods typically rely on singular cell or tissue features and lack design considerations for morphological characteristics of challenging-to-classify categories, resulting in suboptimal classification performance. To address these problems, we proposes a novel breast cancer tissue classification model that Fused features of Edges, Cells, and Tissues (FECT), employing the ResMTUNet and an attention-based aggregator to extract and aggregate these features. Extensive testing on the BRACS dataset demonstrates that our model surpasses current advanced methods in terms of classification accuracy and F1 scores. Moreover, due to its feature fusion that aligns with the diagnostic approach of pathologists, our model exhibits interpretability and holds promise for significant roles in future clinical applications.

Figures

Figures reproduced from arXiv: 2501.10128 by the authors.

Figure 1
Figure 1. Breast cancer tissue image classification framework based on fusion features. a deep learning network capable of simultaneous nucleus segmentation and classification, illustrated at the top branch in [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Edge feature extraction workflow. Similarly, the tissue feature extractor will be trained using the images from the training set and their class labels. 3.4. Edge feature extraction From the cases we have collected, it can be observed that the edges of cancerous tissues are not the typical cord-like structures, making it difficult to discriminate based solely on conventional morphological features. More importantly,… view at source ↗
Figure 3
Figure 3. Comparison of F1 Scores (%) for classifiers based on different features across seven categories of breast cancer tissues [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: The t-SNE dimensionality reduction visualization of cell, tissue, and edge features. Each small dot represents a sample. The fill color of the dot indicates the actual category of the sample, while the outline color of the dot indicates the category to which it has bee…
Figure 4
Figure 4. Figure 4: Impact of different weight combinations of cell and tissue features on classification accuracy and F1 score. The upper heatmap represents accuracy, while the lower one shows the F1 score (%). differences in feature space. Conversely, ADH shows high overlap with UDH and…
Figure 6
Figure 6. Figure 6: Qualitative comparison of the classification effects of tissue features, cell-tissue features and cell-tissue-edge features on difficult-to-classify categories. The classifier pre￾dictions are noted to the right of each example. Red and green indicate incorrect and cor…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 44 canonical work pages

  1. [19]

    Y. Liu, H. Shi, Q. He, Y. Fu, Y. Wang, Y. He, A. Han, T. Guan, Invasive carcinoma segmentation in whole slide images using ms- resmtunet, Heliyon (2024)

  2. [1]

    F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre, A. Jemal, Global cancer statistics 2018: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries, CA: a cancer journal for clinicians 68 (2018) 394–424

  3. [2]

    Allemani, H

    C. Allemani, H. K. Weir, H. Carreira, R. Harewood, D. Spika, X.- S. Wang, F. Bannon, J. V. Ahn, C. J. Johnson, A. Bonaventure, et al., Global surveillance of cancer survival 1995–2009: analysis ofindividualdatafor25676887patientsfrom279population-based registries in 67 countries (concord-2), The lancet 385 (2015) 977– 1010

  4. [3]

    K. Bera, K. A. Schalper, D. L. Rimm, V. Velcheti, A. Madabhushi, Artificial intelligence in digital pathology—new tools for diagnosis and precision oncology, Nature reviews Clinical oncology 16 (2019) 703–715

  5. [4]

    M. K. K. Niazi, A. V. Parwani, M. N. Gurcan, Digital pathology and artificial intelligence, The lancet oncology 20 (2019) e253–e261

  6. [5]

    Campanella, M

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

  7. [6]

    J. N. Kather, A. T. Pearson, N. Halama, D. Jäger, J. Krause, S. H. Loosen, A. Marx, P. Boor, F. Tacke, U. P. Neumann, et al., Deep learning can predict microsatellite instability directly from histology in gastrointestinal cancer, Nature medicine 25 (2019) 1054–1056

  8. [8]

    R. J. Chen, T. Ding, M. Y. Lu, D. F. Williamson, G. Jaume, A. H. Song, B. Chen, A. Zhang, D. Shao, M. Shaban, et al., Towards a general-purpose foundation model for computational pathology, Nature Medicine 30 (2024) 850–862

Show all 51 references
  1. [9]

    M. Y. Lu, B. Chen, D. F. Williamson, R. J. Chen, I. Liang, T. Ding, G. Jaume, I. Odintsov, L. P. Le, G. Gerber, et al., A visual-language foundation model for computational pathology, Nature Medicine 30 (2024) 863–874

  2. [10]

    Y. Zhou, S. Graham, N. Alemi Koohbanani, M. Shaban, P.-A. Heng, N.Rajpoot, Cgc-net:Cellgraphconvolutionalnetworkforgradingof colorectalcancerhistologyimages, in:ProceedingsoftheIEEE/CVF internationalconferenceoncomputervisionworkshops,2019,pp.0– 0

  3. [11]

    P. Pati, G. Jaume, A. Foncubierta-Rodriguez, F. Feroce, A. M. Anni- ciello,G.Scognamiglio,N.Brancati,M.Fiche,E.Dubruc,D.Riccio, etal.,Hierarchicalgraphrepresentationsindigitalpathology,Medical image analysis 75 (2022) 102264

  4. [12]

    Patel, V

    V. Patel, V. Chaurasia, R. Mahadeva, S. P. Patole, Garl-net: Graph based adaptive regularized learning deep network for breast cancer classification, IEEE Access 11 (2023) 9095–9112

  5. [13]

    Z. Shao, H. Bian, Y. Chen, Y. Wang, J. Zhang, X. Ji, et al., Transmil: Transformer based correlated multiple instance learning for whole slideimageclassification, Advancesinneuralinformationprocessing systems 34 (2021) 2136–2147

  6. [14]

    Zhang, C

    J. Zhang, C. Hou, W. Zhu, M. Zhang, Y. Zou, L. Zhang, Q. Zhang, Attentionmultipleinstancelearningwithtransformeraggregationfor breastcancerwholeslideimageclassification, in:2022IEEEInterna- tionalConferenceonBioinformaticsandBiomedicine(BIBM),IEEE, 2022, pp. 1804–1809

  7. [15]

    Stegmüller, B

    T. Stegmüller, B. Bozorgtabar, A. Spahr, J.-P. Thiran, Scorenet: Learning non-uniform attention and augmentation for transformer- based histopathological image classification, in: Proceedings of the IEEE/CVF winter Conference on applications of computer vision, 2023, pp. 6170–6179

  8. [16]

    Dosovitskiy, L

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T.Unterthiner,M.Dehghani,M.Minderer,G.Heigold,S.Gelly,etal., An image is worth 16x16 words: Transformers for image recognition at scale, arXiv preprint arXiv:2010.11929 (2020)

  9. [17]

    K.Wang,F.Zheng,L.Cheng,H.-N.Dai,Q.Dou,J.Qin,Breastcancer classification from digital pathology images via connectivity-aware graph transformer, IEEE Transactions on Medical Imaging (2024)

  10. [18]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez,Ł.Kaiser,I.Polosukhin, Attentionisallyouneed, Advances in neural information processing systems 30 (2017)

  11. [20]

    Brancati, A

    N. Brancati, A. M. Anniciello, P. Pati, D. Riccio, G. Scognamiglio, G.Jaume,G.DePietro,M.DiBonito,A.Foncubierta,G.Botti,etal., Bracs: A dataset for breast carcinoma subtyping in h&e histology images, Database 2022 (2022) baac093

  12. [21]

    Simonyan, A

    K. Simonyan, A. Zisserman, Very deep convolutional networks for large-scale image recognition, arXiv preprint arXiv:1409.1556 (2014)

  13. [22]

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

  14. [23]

    Szegedy, W

    C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, A. Rabinovich, Going deeper with convo- lutions, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1–9

  15. [24]

    1251–1258

    F.Chollet, Xception:Deeplearningwithdepthwiseseparableconvo- lutions, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1251–1258

  16. [25]

    K.Roy,D.Banik,D.Bhattacharjee,M.Nasipuri, Patch-basedsystem for classification of breast histology images using deep learning, Computerized Medical Imaging and Graphics 71 (2019) 90–103

  17. [26]

    F. A. Spanhol, L. S. Oliveira, C. Petitjean, L. Heutte, Breast can- cer histopathological image classification using convolutional neural networks, in: 2016 international joint conference on neural networks (IJCNN), IEEE, 2016, pp. 2560–2567

  18. [27]

    Hameed, B

    Z. Hameed, B. Garcia-Zapirain, J. J. Aguirre, M. A. Isaza-Ruget, Multiclassclassificationofbreastcancerhistopathologyimagesusing multilevel features of deep convolutional neural network, Scientific Reports 12 (2022) 15600

  19. [28]

    T. T. Vuong, B. Song, K. Kim, Y. M. Cho, J. T. Kwak, Multi-scale binarypatternencodingnetworkforcancerclassificationinpathology images, IEEE Journal of Biomedical and Health Informatics 26 (2021) 1152–1163

  20. [29]

    Mehta, X

    S. Mehta, X. Lu, W. Wu, D. Weaver, H. Hajishirzi, J. G. Elmore, L. G. Shapiro, End-to-end diagnosis of breast biopsy images with transformers, Medical image analysis 79 (2022) 102466. Hao et al.: Preprint submitted to Elsevier Page 11 of 12 FECT: Classification of Breast Cance...

  21. [30]

    R. J. Chen, C. Chen, Y. Li, T. Y. Chen, A. D. Trister, R. G. Krishnan, F. Mahmood, Scaling vision transformers to gigapixel images via hi- erarchicalself-supervisedlearning, in:ProceedingsoftheIEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 16144–16155

  22. [31]

    B. Li, Y. Li, K. W. Eliceiri, Dual-stream multiple instance learning network for whole slide image classification with self-supervised contrastive learning, in: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 14318–14328

  23. [32]

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

  24. [33]

    Y.Chen,Z.Shao,H.Bian,Z.Fang,Y.Wang,Y.Cai,H.Wang,G.Liu, X. Li, Y. Zhang, dmil-transformer: Multiple instance learning via integrating morphological and spatial information for lymph node metastasis classification, IEEE Journal of Biomedical and Health Informatics (2023)

  25. [34]

    W. Tang, S. Huang, X. Zhang, F. Zhou, Y. Zhang, B. Liu, Multiple instance learning framework with masked hard instance mining for whole slide image classification, in: Proceedings of the IEEE/CVF InternationalConferenceonComputerVision,2023,pp.4078–4087

  26. [35]

    T. N. Kipf, M. Welling, Semi-supervised classification with graph convolutional networks, arXiv preprint arXiv:1609.02907 (2016)

  27. [36]

    L. Li, Y. Liang, M. Shao, S. Lu, D. Ouyang, et al., Self-supervised learning-based multi-scale feature fusion network for survival anal- ysis from whole slide images, Computers in Biology and Medicine 153 (2023) 106482

  28. [37]

    N. A. Koohbanani, B. Unnikrishnan, S. A. Khurram, P. Krish- naswamy, N. Rajpoot, Self-path: Self-supervision for classification of pathology images with limited annotations, IEEE Transactions on Medical Imaging 40 (2021) 2845–2856

  29. [38]

    C. L. Srinidhi, S. W. Kim, F.-D. Chen, A. L. Martel, Self-supervised driven consistency training for annotation efficient histopathology image analysis, Medical image analysis 75 (2022) 102256

  30. [39]

    O.Ciga,T.Xu,A.L.Martel, Selfsupervisedcontrastivelearningfor digitalhistopathology, MachineLearningwithApplications7(2022) 100198

  31. [40]

    J.Ni,H.Sun,J.Xu,J.Liu,Z.Chen, Afeatureaggregationandfeature fusion network for retinal vessel segmentation, Biomedical Signal Processing and Control 85 (2023) 104829

  32. [41]

    Zhang, H

    L. Zhang, H. Lu, A feature-importance-aware and robust aggregator for gcn, in: Proceedings of the 29th ACM International Conference on Information & Knowledge Management, 2020, pp. 1813–1822

  33. [42]

    Mercan, S

    C. Mercan, S. Aksoy, E. Mercan, L. G. Shapiro, D. L. Weaver, J. G. Elmore, From patch-level to roi-level deep feature representations for breast histopathology classification, in: Medical Imaging 2019: Digital Pathology, volume 10956, SPIE, 2019, pp. 86–93

  34. [43]

    H.Touvron,M.Cord,A.El-Nouby,P.Bojanowski,A.Joulin,G.Syn- naeve,H.Jégou, Augmentingconvolutionalnetworkswithattention- based aggregation, arXiv preprint arXiv:2112.13692 (2021)

  35. [44]

    Y. Cao, H. Ji, W. Zhang, S. Shirani, Feature aggregation networks based on dual attention capsules for visual object tracking, IEEE TransactionsonCircuitsandSystemsforVideoTechnology32(2021) 674–689

  36. [45]

    Graham, Q

    S. Graham, Q. D. Vu, S. E. A. Raza, A. Azam, Y. W. Tsang, J. T. Kwak, N. Rajpoot, Hover-net: Simultaneous segmentation and clas- sification of nuclei in multi-tissue histology images, Medical image analysis 58 (2019) 101563

  37. [46]

    Ronneberger, P

    O. Ronneberger, P. Fischer, T. Brox, U-net: Convolutional networks for biomedical image segmentation, in: Medical image comput- ing and computer-assisted intervention–MICCAI 2015: 18th interna- tionalconference,Munich,Germany,October5-9,2015,proceedings, part III 18, Springer,...

  38. [47]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, L. Fei-Fei, Imagenet: A large-scalehierarchicalimagedatabase, in:2009IEEEconferenceon computer vision and pattern recognition, Ieee, 2009, pp. 248–255

  39. [48]

    14138–14148

    Y.Xiong,Z.Zeng,R.Chakraborty,M.Tan,G.Fung,Y.Li,V.Singh, Nyströmformer: A nyström-based algorithm for approximating self- attention, in: Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 2021, pp. 14138–14148

  40. [49]

    McArt, P

    P.Bankhead,M.B.Loughrey,J.A.Fernández,Y.Dombrowski,D.G. McArt, P. D. Dunne, S. McQuaid, R. T. Gray, L. J. Murray, H. G. Coleman, et al., Qupath: Open source software for digital pathology image analysis, Scientific reports 7 (2017) 1–7

  41. [50]

    Aygüneş, S

    B. Aygüneş, S. Aksoy, R. G. Cinbiş, K. Kösemehmetoğlu, S. Önder, A. Üner, Graph convolutional networks for region of interest classi- fication in breast histopathology, in: Medical Imaging 2020: Digital Pathology, volume 11320, SPIE, 2020, pp. 134–141

  42. [51]

    J. M. J. Valanarasu, V. M. Patel, Unext: Mlp-based rapid medical imagesegmentationnetwork, in:Internationalconferenceonmedical imagecomputingandcomputer-assistedintervention,Springer,2022, pp. 23–33

  43. [52]

    J. Yu, H. Wang, Y. Hao, Two-branchtgnet: A two-branch neural network for breast cancer subtype classification, in: Proceedings of the 2023 4th International Symposium on Artificial Intelligence for Medicine Science, 2023, pp. 1245–1250. Hao et al.: Preprint submitted to Elsevi...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.