Pith. sign in

REVIEW 3 major objections 6 minor 3 cited by

Multimodal Cancer Survival Analysis via Hypergraph Learning with Cross-Modality Rebalance

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read MRePath claims 71.5% mean C-Index for cancer survival prediction by combining sheaf-hypergraph pathology features with dynamic modality rebalancing.

desk verdict Useful incremental method with honest ablations, but the headline C-Index gain is likely inflated by selecting k and the encoder on the same folds used for the final numbers. read the letter →

arxiv 2505.11997 v2 pith:JOWECEQX submitted 2025-05-17 cs.CV

classification cs.CV
keywords survivalpredictionwholeslideimageshypergraphlearningsheafmodalityimbalancemultimodalfusioncancerprognosisC-Index
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 argues that survival prediction from whole-slide pathology images plus genomic profiles is held back by two correctable defects: multi-instance learning aggregates patches without preserving contextual and hierarchical structure, and pathology features numerically dominate genomics during fusion. To fix both, the authors propose MRePath, which models each slide as a hypergraph whose hyperedges connect patches both by spatial proximity and by feature similarity, then propagates information through a sheaf hypergraph Laplacian. A dynamic weighting module estimates each modality's reliability alone and in interaction, and an interactive cross-attention fusion rebalances the two modalities before risk prediction. On five TCGA cancer cohorts, MRePath reports a mean C-Index of 71.5%, about 3 percentage points above the strongest previous multimodal methods compared, PIBD and MOTCat. If correct, the result suggests that higher-order patch relationships and explicit modality rebalancing are worthwhile design choices for prognostic pathology-genomics models.

What carries the argument

The load-bearing machinery is the sheaf hypergraph on the pathology side and the mono/holo-confidence weighting on the fusion side. A hypergraph is a graph in which one hyperedge can connect many nodes at once; here nodes are tissue patches and hyperedges are formed either from spatial proximity in the topological space or from embedding similarity in the feature space, so each hyperedge encodes a local neighbourhood or a recurring morphological pattern. The sheaf hypergraph replaces the ordinary graph Laplacian with a sheaf Laplacian that assigns linear maps between nodes and hyperedges, letting information flow with structure rather than simple averaging, which is what the paper credits for preserving contextual and hierarchical detail. On the fusion side, mono-confidence scores are per-modality reliability estimates produced by MLPs, and holo-confidence is a ratio of log mono-confidences that quantifies how complementary the modalities are; a softmax over their combination produces the final modality weights, and residual co-attention layers then align the reweighted features. The dynamic weighting is presented as a plug-and-play adjustment that could be attached to other multimodal survival architectures.

What would settle it

Re-run the same 5-fold protocol with k selected per training fold, either nested or on a held-out validation split, and compare the mean C-Index against PIBD and MOTCat; if the 3-point gap shrinks to within one standard deviation, the central superiority claim is not supported.

Watch

Extended reading notes

Core claim

The central claim is that replacing patch aggregation with hypergraph learning and counteracting pathology-genomics imbalance yields a multimodal survival model that ranks patient risk more accurately than current multimodal baselines. Concretely, MRePath treats every patch of a whole slide image as a node in a hypergraph built from two kinds of hyperedges: topological ones connecting spatial neighbors within a distance threshold, and feature-based ones connecting patches with similar embeddings. A sheaf hypergraph Laplacian propagates information through these higher-order connections, producing pathology features that retain contextual and hierarchical detail. In parallel, a dynamic weighting scheme computes mono-confidence (per-modality reliability) and holo-confidence (interaction-aware reliability) to reweight pathology and genomic features, and interactive co-attention layers align the two modalities for the final hazard prediction. The reported result is a mean C-Index of 71.5% over BLCA, BRCA, CO-READ, HNSC and STAD, outperforming the second-best compared method by roughly 3 percentage points.

Load-bearing premise

The load-bearing premise is that the hyperedge construction threshold k=9, chosen after inspecting C-Index results on the full benchmark, generalizes to unseen cross-validation folds; if that choice was tuned on the test folds rather than nested within training folds, part of the reported gain may reflect overfitting rather than genuine improvement.

Editorial extensions

If this is right

  • On all five cohorts, MRePath reports the highest C-Index, so the paper claims that hypergraph pathology features combined with rebalanced fusion improve risk ranking over unimodal and multimodal baselines alike.
  • The ablation study reports that combining topological and feature hyperedges (71.5%) beats either type alone (68.7% and 68.8%), implying the two sources of higher-order structure carry complementary prognostic signal.
  • The paper reports that dynamic weighting (71.5%) outperforms fixed weights (67.8%–69.5%) and that interactive alignment fusion beats single-direction cross-attention (66.9%–68.9%), supporting its diagnosis of pathology-genomics imbalance.
  • Kaplan-Meier splits by median predicted risk reach log-rank p<0.05 on all five datasets, which the paper reads as evidence that the model's risk scores stratify patients meaningfully.

Reading between the lines

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

  • A natural extension, hinted at in the limitations, is to apply the mono/holo-confidence weighting to other modality pairs with very different dimensionalities, such as imaging paired with structured clinical records; the mechanism does not depend on the specific biology of pathology or genomics.
  • The contextual-detail claim could be isolated by ablating the hypergraph to a k-nearest-neighbour graph while keeping the same number of neighbours; the authors do not report this comparison, and it would test whether the gains come from higher-order structure or simply from broader local context.
  • The acknowledged equal-k limitation suggests an adaptive per-slide threshold, for example scaling k with patch count or tissue area, as a testable variant; matching or exceeding the reported 71.5% with a data-dependent threshold would weaken concerns that the fixed k=9 was tuned to the benchmark.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes MRePath, a multimodal survival prediction framework that combines sheaf-hypergraph learning on pathology whole-slide images with a dynamic modality-rebalance mechanism for pathology-genomics fusion. The pathology branch constructs topological and feature-based hyperedges over patch-level features and applies a sheaf hypergraph convolution to capture contextual and hierarchical structure; the genomic branch embeds selected gene categories with an MLP. A dynamic weighting scheme computes mono- and holo-confidence scores to reweight the modalities, followed by an interactive alignment fusion with cross-attention. The method is evaluated on five TCGA cancer cohorts (BLCA, BRCA, CO-READ, HNSC, STAD) using 5-fold cross-validation, reporting a mean C-Index of 71.5%, which is 3.0-3.4 percentage points above the best compared baselines (PIBD and MOTCat). The paper also includes ablations on hyperedge types, the threshold k, fusion strategies, and pathology encoders, plus Kaplan-Meier stratification curves.

Significance. If the reported gains are robust, the paper makes a useful empirical contribution: it demonstrates that hypergraph-based patch aggregation and explicit modality reweighting can improve multimodal survival prediction over standard MIL and late-fusion baselines. The authors provide code, evaluate on five public datasets, include per-fold standard deviations, and perform ablations for each proposed module. The Kaplan-Meier analyses with log-rank p-values below 0.05 on all five datasets are also a positive sign of discriminative ability. However, the central claim hinges on whether the reported C-Index advantage is an unbiased estimate or an artifact of model selection on the same folds used for evaluation; this is the main load-bearing concern.

major comments (3)
  1. [Section 4.3, Figure 3, Table 6] The hyperedge construction threshold k is selected after inspecting per-dataset C-Index results on the same 5-fold splits that are subsequently used to report the final numbers. The main text states that increasing k enhances performance with the best results at k=9, and Table 6 in the supplementary shows that k=9 is chosen across all datasets even though other values (e.g., k=48 for CO-READ, k=4 for STAD) sometimes achieve higher means with larger variance. Because no nested cross-validation or held-out selection is used, the reported 71.5% mean C-Index is effectively the maximum over a set of configurations evaluated on the test folds, not an unbiased estimate for a prespecified model. This is load-bearing because the margin over PIBD is only 3.0 points while per-fold standard deviations range from 1.9 to 5.8 points. The same concern applies to the pathology encoder selection in Table 4, where the five encoders produce mean C-Indices from 71.5% to 71.9%, a spread far smaller than the per-fold variability.
  2. [Section 4.2, Table 1] No significance testing is reported for the comparison with baselines. The C-Index tables report mean and standard deviation over five folds, but with only five folds and overlapping patient splits across methods, the 3.0-3.4 point advantage over PIBD/MOTCat may not be statistically reliable. The authors should report paired tests across folds (e.g., Wilcoxon signed-rank or permutation tests) or bootstrap confidence intervals for the per-dataset and mean C-Index differences. Without such tests, the claim that MRePath "consistently outperformed" all methods is not fully supported by the reported uncertainty.
  3. [Section 5, Limitations] The paper acknowledges that a fixed k gives different spatial scopes for slides with very different patch counts, which means the chosen k=9 is data-dependent rather than a scale-invariant default. This limitation interacts with the selection issue: the global threshold is tuned on the benchmark, yet its meaning varies across slides, so the reported generalization may not transfer to cohorts with substantially different WSI sizes or patch densities. The authors should either adapt k per slide based on a principled criterion or evaluate sensitivity across datasets with a proper nested protocol.
minor comments (6)
  1. [Abstract] The abstract claims "over 3.4%" improvement, but Table 1 shows 3.0% over PIBD and 3.4% over MOTCat. The claim should be stated more precisely, e.g., "3.0-3.4 percentage points".
  2. [Section 3.3, Eq. (8)] The holo-confidence formula uses logarithms of wm_p and wm_g without specifying constraints that these values are positive. If mono-confidence outputs are not guaranteed to be strictly positive (e.g., after a linear layer or ReLU), the log is undefined. The authors should clarify the activation and numerical safeguards.
  3. [Section 4.3, Table 4] The text says Conch "emerges as the most effective patch feature encoder," but Table 4 shows mean C-Indices of 71.8 (UNI), 71.9 (Conch), 71.7 (Phikon2), 71.5 (CTransPath), and 71.5 (ResNet50). These differences are negligible relative to the reported standard deviations; the claim should be softened or supported with significance analysis.
  4. [Supplementary, Table 6] The threshold values are inconsistent across the paper: the main text and Figure 3 use k=0,5,9,25,49, while the supplementary Table 6 uses k=0,4,9,24,48. Please align the notation and reported grid.
  5. [Supplementary, Section 8.2] There is a typo: "our resluts" should be "our results."
  6. [Section 4.1] The evaluation section says results are reported as mean C-Index ± STD "across the five datasets," which is ambiguous: it should clarify that the mean is over the five folds within each dataset, and the overall mean is the average of the five per-dataset means.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: MRePath is an empirical architecture paper; self-citations are contextual and not load-bearing.

full rationale

The paper's central claim is Table 1's mean C-Index of 71.5%, obtained by training MRePath on TCGA survival data and evaluating with 5-fold cross-validation (Section 4.1: 'for each cancer type, we conducted 5-fold cross-validation, splitting the data into training and validation sets with a 4:1 ratio'). The survival loss (Eq. 1) is a standard negative log-likelihood supervision on observed events and times, and the dynamic weights in Eqs. 7-9 are learned parameters inside this supervised model, not pre-fit quantities renamed as predictions. The sheaf hypergraph operator is imported from Duta et al. (2024), an external source, and the modality-rebalance confidence measures cite Corbiere et al. (2019) and Cao et al. (2024), also external. Self-citations such as Di et al. (2022a,b) and Jing et al. (2025) appear in related-work comparisons and are used to position the contribution ('in contrast to previous hypergraph studies...'), not to justify the reported gains. No uniqueness theorem or self-citation chain forces the architecture. The main validity concern is model selection: the hyperedge threshold k=9 is chosen after inspecting per-dataset C-Index results on the same folds later used for the headline numbers (Section 4.3, Figure 3, Table 6), and the Limitations section concedes that 'the same k-value in the hypergraph leads to different scopes' for slides with different patch counts. This is a statistical overfitting risk, not a circular derivation: the reported quantity is not definitionally equal to the tuning criterion, and the claim would need held-out validation rather than a circularity correction. Accordingly, no circular step is identified.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central empirical claim depends on standard survival-loss assumptions, on the representation quality of pretrained patch encoders, and on the choice of hyperedge construction threshold k and pathology encoder after inspecting benchmark results. No new physical entities are introduced. The model's learned weights and MLP parameters are ordinary trainable parameters, not independent free parameters.

free parameters (2)
  • Hyperedge construction threshold k = 9
    Selected after comparing k in {0, 5, 9, 25, 49} on all five datasets; controls delta and gamma for topological and feature hyperedges. Performance varies across datasets (Table 6).
  • Pathology feature encoder choice = ResNet50
    The main model uses ResNet50 even though the Conch encoder shows a slightly higher mean C-Index (71.9 vs 71.5) in Table 4; the choice is not justified as part of a proper validation procedure.
assumptions (4)
  • domain assumption Patch-level features from a pretrained encoder are a sufficient representation of pathological morphology for survival prediction.
    Assumed in pathology feature extraction (Section 3.1) and inherited from prior MIL/WSI work.
  • domain assumption Topological and feature-similarity hyperedges with a shared threshold k capture the contextual and hierarchical information that MIL loses.
    Central motivation in Section 3.2; no quantitative verification that hyperedges encode hierarchy beyond downstream C-Index.
  • standard math The discrete negative log-likelihood survival loss (Eq. 1) is a valid objective with the stated event-status convention.
    Relies on the Yao et al. style NLL; the convention ci=0 for event occurrence is unusual and not fully defined.
  • domain assumption TCGA benchmark splits and evaluation protocol from previous studies are appropriate for fair comparison.
    Used in datasets and implementation details; some baselines are cited rather than re-run, so comparability depends on this assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multimodal Cancer Survival Analysis via Hypergraph Learning with Cross-Modality Rebalance." pith.science (2026). https://pith.science/paper/JOWECEQX

@misc{pith2026250511997,
  author       = {Pith},
  title        = {Pith review of: Multimodal Cancer Survival Analysis via Hypergraph Learning with Cross-Modality Rebalance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JOWECEQX}},
  note         = {Machine review of arXiv:2505.11997}
}
read the original abstract

Multimodal pathology-genomic analysis has become increasingly prominent in cancer survival prediction. However, existing studies mainly utilize multi-instance learning to aggregate patch-level features, neglecting the information loss of contextual and hierarchical details within pathology images. Furthermore, the disparity in data granularity and dimensionality between pathology and genomics leads to a significant modality imbalance. The high spatial resolution inherent in pathology data renders it a dominant role while overshadowing genomics in multimodal integration. In this paper, we propose a multimodal survival prediction framework that incorporates hypergraph learning to effectively capture both contextual and hierarchical details from pathology images. Moreover, it employs a modality rebalance mechanism and an interactive alignment fusion strategy to dynamically reweight the contributions of the two modalities, thereby mitigating the pathology-genomics imbalance. Quantitative and qualitative experiments are conducted on five TCGA datasets, demonstrating that our model outperforms advanced methods by over 3.4\% in C-Index performance.

Figures

Figures reproduced from arXiv: 2505.11997 by the authors.

Figure 1
Figure 1. Left: Compared to MIL, hypergraph learning activates more patch regions and better captures contextual and hierarchical details. Right: Examples reveal the pathology-genomics imbalance, where pathology features dominate the overall survival prediction. In multimodal survival analysis, WSIs with giga-level resolution are typically segmented into multiple smaller patches to facilitate computationally efficient analysi… view at source ↗
Figure 2
Figure 2. Overview of MRePath. A, MRePath consists of feature extraction from pathology and genomics modalities, hypergraph learning for capturing WSI representations, and modality rebalance including dynamic weighting and interactive alignment fusion for recalibrating two modalities. B, Hypergraph learning involves constructing topological-based and feature-based hyperedges, and employing sheaf hypergraph to encourage local … view at source ↗
Figure 3
Figure 3. Ablation study on hyperedge construction threshold k. Performance of various similarity thresholds (k = 0, 5, 9, 25, 49) is reported in C-Index (%). The baseline model without hyperedges (k = 0) showed mod￾erate performance across all five datasets. Increasing k en￾hanced performance, with the best results achieved at k = 9. However, further increasing k led to a decline in performance and more computational costs. … view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Visualization for low and high-risk cases in the BRCA (Top): The heatmaps are generated using cross-attention scores, with red and blue for high and low scores. The top five most influential genes are also highlighted in red for high and blue for low. Kaplan-Meier curv…
Figure 5
Figure 5. Figure 5: Visualization for the distribution of each dataset using box [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: a), Density of modality weights, where purple indicates WSI (pathology) and green indicates genomics. b), Visualization of the prediction layer weights after applying modality rebalance. formance across all models, with Conch achieving the high￾est mean score of 83.2% …
Figure 7
Figure 7. Figure 7: For visualization of the BLCA, CO-READ, HNSC and STAD dataset, one low-risk patient (Top) and one high-risk patient (Bottom) [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. FraPPE: Fast and Efficient Preference-based Pure Exploration

    cs.LG 2025-08 unverdicted novelty 6.0 of 10

    The abstract claims a sample-optimal, O(KL^2) preference exploration algorithm, but the supplied full text is an unrelated medical imaging manuscript.

  2. Together, Then Apart: Balancing Alignment and Distinctiveness for Multimodal Survival Analysis

    cs.CV 2025-11 conditional novelty 5.0 of 10

    TTA reports improved C-index on five TCGA cohorts by aligning histology and genomics to shared prototypes with unbalanced optimal transport, then enforcing modality distinctiveness with anchor-based contrastive regula...

  3. Memory-Augmented Incomplete Multimodal Survival Prediction via Cross-Slide and Gene-Attentive Hypergraph Learning

    cs.CV 2025-06 conditional novelty 5.0 of 10

    M2Surv reports higher concordance index than prior multimodal survival models on five TCGA datasets by adding hypergraph-based multi-slide fusion, gene-attention hyperedges, and a memory bank for missing-modality inference.

Reference graph

Works this paper leans on

46 extracted references · 43 canonical work pages · cited by 3 Pith papers

  1. [1]

    Representation learning of histopathology images using graph neural networks

    [Adnan et al., 2020] Mohammed Adnan, Shivam Kalra, and Hamid R Tizhoosh. Representation learning of histopathology images using graph neural networks. In CVPR Workshops, pages 988–989,

  2. [7]

    Pathomic fusion: an integrated framework for fusing histopathology and genomic features for cancer diagnosis and prognosis

    [Chen et al., 2020] Richard J Chen, Ming Y Lu, Jingwen Wang, et al. Pathomic fusion: an integrated framework for fusing histopathology and genomic features for cancer diagnosis and prognosis. IEEE Transactions on Medical Imaging, 41(4):757–770,

  3. [9]

    Towards a general-purpose foundation model for computational pathology

    [Chen et al., 2024] Richard J Chen, Tong Ding, Ming Y Lu, Drew FK Williamson, et al. Towards a general-purpose foundation model for computational pathology. Nature Medicine, 30(3):850–862,

  4. [10]

    Ad- dressing failure prediction by learning model confidence

    [Corbi`ere et al., 2019] Charles Corbi `ere, Nicolas Thome, Avner Bar-Hen, Matthieu Cord, and Patrick P ´erez. Ad- dressing failure prediction by learning model confidence. NeurIPS, 32,

  5. [11]

    Solving the multiple instance problem with axis-parallel rectangles

    [Dietterich et al., 1997] Thomas G Dietterich, Richard H Lathrop, and Tom ´as Lozano-P ´erez. Solving the multiple instance problem with axis-parallel rectangles. Artificial intelligence, 89(1-2):31–71,

  6. [13]

    Learning visual features by colorization for slide-consistent survival prediction from whole slide images

    [Fan et al., 2021] Lei Fan, Arcot Sowmya, Erik Meijering, and Yang Song. Learning visual features by colorization for slide-consistent survival prediction from whole slide images. In MICCAI, pages 592–601. Springer,

  7. [14]

    Hypergraph neural net- works

    [Feng et al., 2019] Yifan Feng, Haoxuan You, Zizhao Zhang, Rongrong Ji, and Yue Gao. Hypergraph neural net- works. In AAAI, volume 33, pages 3558–3565,

  8. [15]

    Scaling self-supervised learning for histopathology with masked image modeling

    [Filiot et al., 2023] Alexandre Filiot, Ridouane Ghermi, An- toine Olivier, Paul Jacob, et al. Scaling self-supervised learning for histopathology with masked image modeling. medRxiv,

Show all 46 references
  1. [16]

    Node-aligned graph convolutional network for whole-slide image representation and classification

    [Guan et al., 2022] Yonghang Guan, Jun Zhang, Kuan Tian, et al. Node-aligned graph convolutional network for whole-slide image representation and classification. In CVPR, pages 18813–18823,

  2. [17]

    Neural networks: a compre- hensive foundation

    [Haykin, 1998] Simon Haykin. Neural networks: a compre- hensive foundation. Prentice Hall PTR,

  3. [20]

    Hˆ 2-mil: exploring hierarchical representation with heterogeneous multiple instance learning for whole slide image analysis

    [Hou et al., 2022] Wentai Hou, Lequan Yu, Chengxuan Lin, et al. Hˆ 2-mil: exploring hierarchical representation with heterogeneous multiple instance learning for whole slide image analysis. In AAAI, volume 36, pages 933–941,

  4. [21]

    Attention-based deep multiple instance learning

    [Ilse et al., 2018] Maximilian Ilse, Jakub Tomczak, and Max Welling. Attention-based deep multiple instance learning. In ICML, pages 2127–2136. PMLR,

  5. [23]

    Hmil: Hierarchical multi-instance learning for fine- grained whole slide image classification

    [Jin et al., 2024] Cheng Jin, Luyang Luo, Huangjing Lin, et al. Hmil: Hierarchical multi-instance learning for fine- grained whole slide image classification. IEEE Transac- tions on Medical Imaging,

  6. [24]

    Multi-modal hy- pergraph contrastive learning for medical image segmen- tation

    [Jing et al., 2025] Weipeng Jing, Junze Wang, Donglin Di, Dandan Li, Yang Song, and Lei Fan. Multi-modal hy- pergraph contrastive learning for medical image segmen- tation. Pattern Recognition, 165:111544,

  7. [26]

    Self- normalizing neural networks

    [Klambauer et al., 2017] G¨unter Klambauer, Thomas Un- terthiner, Andreas Mayr, and Sepp Hochreiter. Self- normalizing neural networks. NeurIPS, 30,

  8. [28]

    High-order correlation-guided slide-level histology retrieval with self-supervised hash- ing

    [Li et al., 2023] Shengrui Li, Yining Zhao, Jun Zhang, Ting Yu, Ji Zhang, and Yue Gao. High-order correlation-guided slide-level histology retrieval with self-supervised hash- ing. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(9):11008–11023,

  9. [29]

    Artificial intelligence for multimodal data integration in oncology

    [Lipkova et al., 2022] Jana Lipkova, Richard J Chen, Bowen Chen, et al. Artificial intelligence for multimodal data integration in oncology. Cancer cell, 40(10):1095–1110,

  10. [30]

    Data-efficient and weakly super- vised computational pathology on whole-slide images

    [Lu et al., 2021] Ming Y Lu, Drew FK Williamson, Tiffany Y Chen, et al. Data-efficient and weakly super- vised computational pathology on whole-slide images. Nature biomedical engineering, 5(6):555–570,

  11. [31]

    A visual-language foundation model for computational pathology

    [Lu et al., 2024] Ming Y Lu, Bowen Chen, Drew FK Williamson, et al. A visual-language foundation model for computational pathology. Nature Medicine , 30(3):863– 874,

  12. [32]

    Predicting cancer out- comes from histology and genomics using convolutional networks

    [Mobadersany et al., 2018] Pooya Mobadersany, Safoora Yousefi, Mohamed Amgad, et al. Predicting cancer out- comes from histology and genomics using convolutional networks. Proceedings of the National Academy of Sci- ences, 115(13):E2970–E2979,

  13. [33]

    Sparse multi-modal graph transformer with shared-context processing for representation learning of giga-pixel images

    [Nakhli et al., 2023] Ramin Nakhli, Puria Azadi Moghadam, Haoyang Mi, et al. Sparse multi-modal graph transformer with shared-context processing for representation learning of giga-pixel images. InCVPR, pages 11547–11557,

  14. [34]

    Prognostic genome and transcriptome signatures in colorectal cancers

    [Nunes et al., 2024] Lu´ıs Nunes, Fuqiang Li, Meizhen Wu, et al. Prognostic genome and transcriptome signatures in colorectal cancers. Nature, 633(8028):137–146,

  15. [35]

    Boundary-guided learning for gene expression prediction in spatial transcriptomics

    [Qu et al., 2024] Mingcheng Qu, Yuncong Wu, Donglin Di, Anyang Su, Tonghua Su, Yang Song, and Lei Fan. Boundary-guided learning for gene expression prediction in spatial transcriptomics. In 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) , pages 445...

  16. [36]

    Noise in gene expression: origins, consequences, and control

    [Raser and O’shea, 2005] Jonathan M Raser and Erin K O’shea. Noise in gene expression: origins, consequences, and control. Science, 309(5743):2010–2013,

  17. [38]

    Tumor micro- environment interactions guided graph learning for sur- vival analysis of human cancers from whole-slide patho- logical images

    [Shao et al., 2024] Wei Shao, YangYang Shi, Daoqiang Zhang, JunJie Zhou, and Peng Wan. Tumor micro- environment interactions guided graph learning for sur- vival analysis of human cancers from whole-slide patho- logical images. In CVPR, pages 11694–11703,

  18. [39]

    Prototype-based image prompting for weakly supervised histopathological image segmenta- tion

    [Tang et al., 2025] Qingchen Tang, Lei Fan, Maurice Pag- nucco, and Yang Song. Prototype-based image prompting for weakly supervised histopathological image segmenta- tion. arXiv preprint arXiv:2503.12068,

  19. [40]

    Transformer-based unsupervised con- trastive learning for histopathological image classification

    [Wang et al., 2022] Xiyue Wang, Sen Yang, Jun Zhang, Minghui Wang, Jing Zhang, Wei Yang, Junzhou Huang, and Xiao Han. Transformer-based unsupervised con- trastive learning for histopathological image classification. Medical image analysis, 81:102559,

  20. [41]

    Multi- modal optimal transport-based co-attention transformer with global structure consistency for survival prediction

    [Xu and Chen, 2023] Yingxue Xu and Hao Chen. Multi- modal optimal transport-based co-attention transformer with global structure consistency for survival prediction. In ICCV, pages 21241–21251,

  21. [42]

    Whole slide images based cancer survival prediction using at- tention guided deep multiple instance learning networks

    [Yao et al., 2020] Jiawen Yao, Xinliang Zhu, Jitendra Jonna- gaddala, Nicholas Hawkins, and Junzhou Huang. Whole slide images based cancer survival prediction using at- tention guided deep multiple instance learning networks. Medical Image Analysis, 65:101789,

  22. [43]

    [Yao et al., 2021] Jiawen Yao, Yu Shi, Kai Cao, et al. Deep- prognosis: Preoperative prediction of pancreatic cancer survival and surgical margin via comprehensive under- standing of dynamic contrast-enhanced ct imaging and tumor-vascular contact parsing. Medical image analysi...

  23. [44]

    Prototypical information bottlenecking and disentangling for multimodal cancer survival prediction

    [Zhang et al., 2024] Yilan Zhang, Yingxue Xu, Jianqi Chen, Fengying Xie, and Hao Chen. Prototypical information bottlenecking and disentangling for multimodal cancer survival prediction. In ICLR,

  24. [45]

    Histopathological whole slide image analysis using context-based cbir

    [Zheng et al., 2018] Yushan Zheng, Zhiguo Jiang, Haopeng Zhang, et al. Histopathological whole slide image analysis using context-based cbir. IEEE transactions on medical imaging, 37(7):1641–1652,

  25. [46]

    Cross- modal translation and alignment for survival analysis

    [Zhou and Chen, 2023] Fengtao Zhou and Hao Chen. Cross- modal translation and alignment for survival analysis. In ICCV, pages 21485–21494,

  26. [1997]

    Sheaf hypergraph networks

    [Duta et al., 2024] Iulia Duta, Giulia Cassar `a, Fabrizio Sil- vestri, and Pietro Li `o. Sheaf hypergraph networks. NeurIPS, 36,

  27. [1998]

    Deep residual learning for image recog- nition

    [He et al., 2016] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recog- nition. In CVPR, pages 770–778,

  28. [2005]

    Transmil: Transformer based correlated multi- ple instance learning for whole slide image classification

    [Shao et al., 2021] Zhuchen Shao, Hao Bian, Yang Chen, et al. Transmil: Transformer based correlated multi- ple instance learning for whole slide image classification. NeurIPS, 34:2136–2147,

  29. [2016]

    Patch- based convolutional neural network for whole slide tissue image classification

    [Hou et al., 2016] Le Hou, Dimitris Samaras, Tahsin M Kurc, Yi Gao, James E Davis, and Joel H Saltz. Patch- based convolutional neural network for whole slide tissue image classification. In CVPR, pages 2424–2433,

  30. [2017]

    Dual- stream multiple instance learning network for whole slide image classification with self-supervised contrastive learn- ing

    [Li et al., 2021] Bin Li, Yin Li, and Kevin W Eliceiri. Dual- stream multiple instance learning network for whole slide image classification with self-supervised contrastive learn- ing. In CVPR, pages 14318–14328,

  31. [2018]

    Modeling dense multimodal in- teractions between biological pathways and histology for survival prediction

    [Jaume et al., 2024] Guillaume Jaume, Anurag Vaidya, Richard J Chen, et al. Modeling dense multimodal in- teractions between biological pathways and histology for survival prediction. In CVPR, pages 11579–11590,

  32. [2019]

    Predictive dynamic fusion

    [Cao et al., 2024] Bing Cao, Yinan Xia, Yi Ding, Changqing Zhang, and Qinghua Hu. Predictive dynamic fusion. In ICML,

  33. [2020]

    Genomics to select treatment for patients with metastatic breast cancer.Nature, 610(7931):343–348,

    [Andre et al., 2022] Fabrice Andre, Thomas Filleron, Maud Kamal, et al. Genomics to select treatment for patients with metastatic breast cancer.Nature, 610(7931):343–348,

  34. [2021]

    Pan-cancer integrative histology- genomic analysis via multimodal deep learning

    [Chen et al., 2022] Richard J Chen, Ming Y Lu, Drew FK Williamson, et al. Pan-cancer integrative histology- genomic analysis via multimodal deep learning. Cancer Cell, 40(8):865–878,

  35. [2022]

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

    [Campanella et al., 2019] Gabriele Campanella, Matthew G Hanna, Luke Geneslaw, et al. Clinical-grade computa- tional pathology using weakly supervised deep learning on whole slide images. Nature medicine, 25(8):1301–1309,

  36. [2023]

    Deep learning with multimodal representation for pancancer prognosis prediction

    [Cheerla and Gevaert, 2019] Anika Cheerla and Olivier Gevaert. Deep learning with multimodal representation for pancancer prognosis prediction. Bioinformatics, 35(14):i446–i454,

  37. [2024]

    Histopathology whole slide image analysis with heterogeneous graph rep- resentation learning

    [Chan et al., 2023] Tsai Hor Chan, Fernando Julio Cendra, Lan Ma, Guosheng Yin, and Lequan Yu. Histopathology whole slide image analysis with heterogeneous graph rep- resentation learning. In CVPR, pages 15661–15670,

  38. [2025]

    Si-mil: Taming deep mil for self-interpretability in gigapixel histopathology

    [Kapse et al., 2024] Saarthak Kapse, Pushpak Pati, Srijan Das, et al. Si-mil: Taming deep mil for self-interpretability in gigapixel histopathology. In CVPR, pages 11226– 11237,

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.