pith. sign in

arxiv: 2602.05126 · v3 · pith:PCN5GDEOnew · submitted 2026-02-04 · 💻 cs.CV

CLEAR-HPV: Interpretable concept discovery for human-papillomavirus-associated morphology in whole-slide histology

Weiyi Qin , Yingci Liu-Swetz , Shiwei Tan , Hao Wang This is my paper

Pith reviewed 2026-05-25 07:16 UTC · model grok-4.3

classification 💻 cs.CV
keywords HPVhistopathologyconcept discoverymultiple instance learningwhole-slide imaginginterpretable AIattention mechanismsmorphologic concepts
0
0 comments X

The pith

CLEAR-HPV restructures attention-based MIL latent space to discover keratinizing, basaloid and stromal concepts for HPV prediction without concept labels.

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

The paper establishes that re-weighting the latent space of attention-based multiple instance learning models with attention weights enables automatic discovery of morphologic concepts tied to HPV status. These concepts appear as keratinizing, basaloid, and stromal patterns, which can be visualized as spatial maps on whole-slide images and summarized as compact concept-fraction vectors per slide. The vectors compress embeddings from 1536 dimensions down to 10 while retaining the original predictive performance for HPV classification. A sympathetic reader would care because this supplies visual and numerical interpretability to otherwise opaque slide-level predictions that guide prognosis and treatment in head and neck and cervical cancers. The method works consistently on TCGA-HNSCC, TCGA-CESC, and CPTAC-HNSCC without requiring concept supervision during training and remains agnostic to the choice of MIL backbone.

Core claim

CLEAR-HPV restructures the MIL latent space using attention to enable concept discovery without requiring concept labels during training. Operating in an attention-weighted latent space, CLEAR-HPV automatically discovers keratinizing, basaloid, and stromal morphologic concepts, generates spatial concept maps, and represents each slide using a compact concept-fraction vector. CLEAR-HPV's concept-fraction vectors preserve the predictive information of the original MIL embeddings while reducing the high-dimensional feature space to only 10 interpretable concepts. CLEAR-HPV generalizes consistently across TCGA-HNSCC, TCGA-CESC, and CPTAC-HNSCC.

What carries the argument

Attention-weighted latent space restructuring that isolates morphologic concepts and produces concept-fraction vectors from attention-based MIL embeddings.

If this is right

  • Each slide is represented by a 10-dimensional concept-fraction vector that matches the HPV prediction accuracy of the original high-dimensional embedding.
  • Spatial concept maps highlight the locations of keratinizing, basaloid, and stromal patterns within each whole-slide image.
  • The framework applies across different attention-based MIL backbones and three independent datasets without retraining for concept labels.
  • High-dimensional feature spaces can be replaced by these compact interpretable vectors for downstream analysis while keeping predictive power.

Where Pith is reading between the lines

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

  • The same attention-restructuring step could be tested on other slide-level biomarkers such as PD-L1 or microsatellite instability to check whether similar concept discovery occurs.
  • Concept-fraction vectors might serve as input features for survival models to test whether the discovered morphology concepts carry prognostic value beyond HPV status alone.
  • If the concepts align with known histologic subtypes, the method could support prospective studies that measure whether providing these maps to pathologists changes diagnostic consistency.

Load-bearing premise

That attention weights in the latent space isolate distinct morphologic patterns that correspond to actual tissue biology without any concept supervision.

What would settle it

Pathologist manual annotation of keratinizing, basaloid, and stromal regions on a held-out set of slides; if the automatically generated spatial concept maps show low overlap with those annotations, the discovery claim fails.

Figures

Figures reproduced from arXiv: 2602.05126 by Hao Wang, Shiwei Tan, Weiyi Qin, Yingci Liu-Swetz.

Figure 1
Figure 1. Figure 1: Overview of the CLEAR-HPV framework. (A) Data processing pipeline: WSIs are de￾composed into fixed-size tiles, encoded with a pretrained ViT or CNN, and converted into patch￾level feature embeddings. (B) An attention-based MIL classifier projects embeddings into the h-space latent representation and uses multi-head attention to compute tile-level contributions, which are pooled into a single slide-level em… view at source ↗
Figure 2
Figure 2. Figure 2: Recovery score relative to the interpreted MIL model (CLAM) across ACC, AUC, F1, Precision, Recall (i.e., sensitivity), and Specificity. For each method, the Euclidean distance d between its metric vector (i.e., concatenation of Ac￾curacy, AUC, etc.) and the interpreted model’s is computed and converted to a similarity score s = 1 1+d . Higher scores in￾dicate closer agreement with CLAM [PITH_FULL_IMAGE:f… view at source ↗
Figure 3
Figure 3. Figure 3: Class-averaged concept-fraction vectors across concept-discovery settings on TCGA-HNSCC. Concept-fraction vectors are computed per slide as the fraction of tiles assigned to each discovered concept, optionally weighted by MIL attention. These slide-level vectors are then averaged within each group to obtain class-averaged profiles that summarize cohort-level morphologic composition and highlight difference… view at source ↗
Figure 4
Figure 4. Figure 4: Top tiles for key concepts discovered by CLEAR-HPV (A) and the corresponding slide-level distributions in the dataset TCGA-HNSCC (B). (A) Top (representative) tiles for five CLEAR-HPV concepts chosen for their consistent appearance and clear morphologic identity: C5 (basaloid squamous epithelium), C7 (keratinizing squamous epithelium), C9 (fibrous stroma), C4 (connective stroma), and C2 (inflammatory cells… view at source ↗
Figure 5
Figure 5. Figure 5: Visualization of attention-weighted concept discovery using CLEAR-HPV. (A) For representative HPV-positive and HPV-negative WSIs from TCGA-HNSCC, we show, in four columns: (i) the original H&E whole slide image, (ii) the h-space spatial concept map, (iii) the high-attention spatial concept map, and (iv) regions of interest (ROIs) with their correspond￾ing concept-fraction distributions produced by our CLEA… view at source ↗
Figure 6
Figure 6. Figure 6: Cross-cohort consistency of HPV-related concepts among top-8 tiles. We show representative high-attention tiles for the HPV-positive-related “basaloid” concept C5 and the “keratinizing” concept C7 from two external cohorts, TCGA-CESC (top) and CPTAC-HNSCC (bottom). Across both datasets, C5 consistently reflects basaloid morphology characteris￾tic of HPV-positive tumors, while C7 reflects keratinizing morph… view at source ↗
read the original abstract

Human papillomavirus (HPV) status is a critical determinant of prognosis and treatment response in head and neck and cervical cancers. Although attention-based multiple instance learning (MIL) achieves strong slide-level prediction for HPV-related whole-slide histopathology, it provides limited morphologic interpretability. To address this limitation, we introduce Concept-Level Explainable Attention-guided Representation for HPV (CLEAR-HPV), a framework that restructures the MIL latent space using attention to enable concept discovery without requiring concept labels during training. Operating in an attention-weighted latent space, CLEAR-HPV automatically discovers keratinizing, basaloid, and stromal morphologic concepts, generates spatial concept maps, and represents each slide using a compact concept-fraction vector. CLEAR-HPV's concept-fraction vectors preserve the predictive information of the original MIL embeddings while reducing the high-dimensional feature space (e.g., 1536 dimensions) to only 10 interpretable concepts. CLEAR-HPV generalizes consistently across TCGA-HNSCC, TCGA-CESC, and CPTAC-HNSCC, providing compact, concept-level interpretability through a general, backbone-agnostic framework for attention-based MIL models of whole-slide histopathology.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 2 minor

Summary. The manuscript introduces CLEAR-HPV, a framework that restructures the latent space of attention-based multiple instance learning (MIL) models to enable unsupervised discovery of morphologic concepts (keratinizing, basaloid, stromal) in HPV-associated whole-slide histopathology. It generates spatial concept maps and represents slides with 10-dimensional concept-fraction vectors that preserve the predictive power of the original high-dimensional (e.g., 1536) MIL embeddings while generalizing across TCGA-HNSCC, TCGA-CESC, and CPTAC-HNSCC cohorts.

Significance. If the quantitative results support the claims, this provides a general, backbone-agnostic method for adding concept-level interpretability to MIL models in computational pathology without requiring labeled concepts during training. The reduction to a compact, interpretable representation while maintaining predictive performance and cross-dataset consistency could facilitate clinical translation by linking model predictions to known histologic features. The unsupervised concept discovery and reported generalization across independent cohorts are notable strengths.

minor comments (2)
  1. [Abstract] Abstract: The abstract asserts that concept-fraction vectors 'preserve the predictive information' and 'generalize consistently' but supplies no numerical metrics (e.g., AUC deltas, correlation coefficients, or ablation results); a brief quantitative summary should be added.
  2. [Methods] The description of how the attention-weighted latent space is restructured for concept discovery (e.g., clustering or factorization procedure) is referenced but not detailed in the provided abstract; ensure the methods section supplies the exact algorithm and hyperparameters.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of CLEAR-HPV, the recognition of its backbone-agnostic interpretability contribution, and the recommendation for minor revision. No specific major comments appear in the provided report, so we have no point-by-point replies to offer. We remain ready to incorporate any additional feedback or perform minor edits if requested by the editor.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents CLEAR-HPV as a restructuring of attention-based MIL latent space to enable unsupervised concept discovery, with claims that concept-fraction vectors preserve predictive information while reducing dimensionality from 1536 to 10 concepts. No equations, derivations, or self-citations are provided in the abstract or description that would reduce any prediction or result to fitted inputs by construction. The framework is described as backbone-agnostic and generalizing across independent cohorts (TCGA-HNSCC, TCGA-CESC, CPTAC-HNSCC), with no load-bearing steps that equate outputs to inputs via definition or self-referential fitting. The derivation chain appears self-contained as an extension of existing MIL methods without circular reductions.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no information on free parameters, axioms, or invented entities. The described approach appears to rely on standard attention mechanisms in MIL but specifics are unavailable.

pith-pipeline@v0.9.0 · 5748 in / 1277 out tokens · 53530 ms · 2026-05-25T07:16:51.377123+00:00 · methodology

Review history (2 revisions) →

discussion (0)

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

Reference graph

Works this paper leans on

45 extracted references · 45 canonical work pages · 3 internal anchors

  1. [1]

    Global burden of cancer attributable to infections in 2018: a worldwide incidence analysis

    Catherine de Martel, Damien Georges, Freddie Bray, Jacques Ferlay, and Gary M Clifford. Global burden of cancer attributable to infections in 2018: a worldwide incidence analysis. The Lancet global health, 8(2):180–190, 2020

    work page 2018

  2. [2]

    Ang, Jay Harris, Robert Wheeler, Randal Weber, David I

    Kevin K. Ang, Jay Harris, Robert Wheeler, Randal Weber, David I. Rosenthal, P . F . Nguyen- Tˆan, William H. Westra, Christine H. Chung, R. Clay Jordan, Chaomei Lu, Hyejin Kim, Rebecca Axelrod, Carole C. Silverman, Kimberly P . Redmond, and Maura L. Gillison. Hu- man papillomavirus and survival of patients with oropharyngeal cancer.The New England Journal...

    work page doi:10.1056/nejmoa0912217 2010

  3. [3]

    Gillison, William M

    Maura L. Gillison, William M. Koch, Ralph B. Capone, Michael Spafford, William H. Westra, Ling Wu, Mark L. Zahurak, Robert W. Daniel, Michael Viglione, Daniel E. Symer, Keerti V. Shah, and David Sidransky. Evidence for a causal association between human papillo- mavirus and a subset of head and neck cancers.Journal of the National Cancer Institute, 92(9):...

    work page doi:10.1093/jnci/92.9.709 2000

  4. [4]

    Marur and A

    S. Marur and A. A. Forastiere. Head and neck cancer: changing epidemiology, diagnosis, and treatment.Mayo Clinic Proceedings, 83(4):489–501, April 2008. doi:10.4065/83.4.489. Erratum in Mayo Clin Proc. 2008 May;83(5):604

    work page doi:10.4065/83.4.489 2008

  5. [5]

    O. R. Linton, M. G. Moore, J. S. Brigance, C. A. Gordon, D. J. Summerlin, and M. W. McDonald. Prognostic significance of basaloid squamous cell carcinoma in head and neck cancer.JAMA Otolaryngology–Head & Neck Surgery, 139(12):1306–1311, 2013. doi:10. 1001/jamaoto.2013.5308

    work page arXiv 2013

  6. [6]

    A. A. Shah, S. K. Jeffus, and E. B. Stelow. Squamous cell carcinoma variants of the upper aerodigestive tract: a comprehensive review with a focus on genetic alter- ations.Archives of Pathology & Laboratory Medicine, 138(6):731–744, June 2014. doi: 10.5858/arpa.2013-0070-RA

    work page doi:10.5858/arpa.2013-0070-ra 2014

  7. [7]

    Deep learning can predict microsatellite instability directly from histology in gastroin- testinal cancer.Nature Medicine, 25:1054–1056, 2019

    Jakob N Kather, Alexander T Pearson, Niels Halama, Dirk J ¨ager, Jos´e Krause, Stefanie H Loosen, Alexander Marx, Sabine Foersch, Ansgar D¨usterhoft, Thomas van den Heuvel, and et al. Deep learning can predict microsatellite instability directly from histology in gastroin- testinal cancer.Nature Medicine, 25:1054–1056, 2019. doi:10.1038/s41591-019-0462-y

    work page doi:10.1038/s41591-019-0462-y 2019

  8. [8]

    J. S. Jr Lewis, B. Beadle, J. A. Bishop, R. D. Chernock, C. Colasacco, C. Lacchetti, J. T. Moncur, J. W. Rocco, M. R. Schwartz, R. R. Seethala, N. E. Thomas, W. H. Westra, and W. C. Faquin. Human papillomavirus testing in head and neck carcinomas: Guideline from the college of american pathologists.Archives of Pathology & Laboratory Medicine, 142(5): 559–...

    work page doi:10.5858/arpa.2017-0286-cp 2018

  9. [9]

    Hanna, Lisa Geneslaw, Albert Miraflor, Victor Werneck Krauss Silva, Klaus J

    Giacomo Campanella, Michael G. Hanna, Lisa Geneslaw, Albert Miraflor, Victor Werneck Krauss Silva, Klaus J. Busam, Edi Brogi, Victor E. Reuter, David S. Klimstra, and Thomas J. Fuchs. Clinical-grade computational pathology using weakly supervised deep learning on whole slide images.Nature Medicine, 25:1301–1309, 2019. doi:10.1038/s41591-019-0508-1

    work page doi:10.1038/s41591-019-0508-1 2019

  10. [10]

    Ocampo, Theodore Sakellaropoulos, Navneet Narula, Matija Snuderl, David Feny¨o, Andre L

    Nicolas Coudray, Paola S. Ocampo, Theodore Sakellaropoulos, Navneet Narula, Matija Snuderl, David Feny¨o, Andre L. Moreira, Narges Razavian, and Aristotelis Tsirigos. Classifi- cation and mutation prediction from non–small cell lung cancer histopathology images using deep learning.Nature Medicine, 24:1559–1567, 2018. doi:10.1038/s41591-018-0177-5. 22

    work page doi:10.1038/s41591-018-0177-5 2018

  11. [11]

    Y . Liu, T. Kohlberger, M. Norouzi, G. E. Dahl, L. Metz, J. Shoemaker, R. MacDonald, C. Scott, P . Q. Nelson, J. D. Hipp, G. S. Corrado, J. Dean, H. MacMahon, A. Madab- hushi, and M. C. Stumpe. Artificial intelligence–based breast cancer nodal metastasis de- tection: Insights into the black box for pathologists.JAMA, 323(4):315–325, 2020. doi: 10.1001/jam...

    work page doi:10.1001/jama.2019.2625 2020

  12. [12]

    Explainable artificial intel- ligence: Understanding, visualizing and interpreting deep learning models.IEEE Signal Processing Magazine, 34(6):26–41, 2017

    Wojciech Samek, Thomas Wiegand, and Klaus-Robert M ¨uller. Explainable artificial intel- ligence: Understanding, visualizing and interpreting deep learning models.IEEE Signal Processing Magazine, 34(6):26–41, 2017

    work page 2017

  13. [13]

    Explainable artificial intelligence (xai): Concepts, taxonomies, opportunities and challenges toward responsible ai.Information Fusion, 58:82–115, 2020

    Alejandro Barredo Arrieta, Natalia D ´ıaz-Rodr´ıguez, Javier Del Ser, Adrien Bennetot, Siham Tabik, Alberto Barbado, Salvador Garc ´ıa, Sergio Gil-L´opez, Daniel Molina, Richard Ben- jamins, et al. Explainable artificial intelligence (xai): Concepts, taxonomies, opportunities and challenges toward responsible ai.Information Fusion, 58:82–115, 2020

    work page 2020

  14. [14]

    A survey on explainable artificial intelligence (xai): Toward medical xai.Computer Methods and Programs in Biomedicine, 200:105009, 2021

    Erico Tjoa and Cuntai Guan. A survey on explainable artificial intelligence (xai): Toward medical xai.Computer Methods and Programs in Biomedicine, 200:105009, 2021

    work page 2021

  15. [15]

    Attention-based Deep Multiple Instance Learning

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

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  16. [16]

    Lu, Drew F

    Ming Y . Lu, Drew F . K. Williamson, Tiffany Y . Chen, Rui Chen, Mauro Barbieri, and Faisal Mahmood. Data-efficient and weakly supervised computational pathology on whole-slide images.Nature Biomedical Engineering, 5:555–570, 2021. doi:10.1038/ s41551-020-00682-w

    work page 2021

  17. [17]

    Trans- mil: Transformer based correlated multiple instance learning for whole slide image classifi- cation.Advances in Neural Information Processing Systems, 34:2136–2147, 2021

    Zhuchen Shao, Hao Bian, Y ang Chen, Yifeng Wang, Jian Zhang, Xiangyang Ji, et al. Trans- mil: Transformer based correlated multiple instance learning for whole slide image classifi- cation.Advances in Neural Information Processing Systems, 34:2136–2147, 2021

    work page 2021

  18. [18]

    Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, and Dhruv Batra

    Ramprasaath R. Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, and Dhruv Batra. Grad-cam: Visual explanations from deep networks via gradient- based localization.International Journal of Computer Vision, 128(2):336–359, October

    work page

  19. [19]

    doi:10.1007/s11263-019-01228-7

    ISSN 1573-1405. doi:10.1007/s11263-019-01228-7. URLhttp://dx.doi.org/10. 1007/s11263-019-01228-7

    work page doi:10.1007/s11263-019-01228-7

  20. [20]

    Representation Learning: A Review and New Perspectives

    Y oshua Bengio, Aaron Courville, and Pascal Vincent. Representation learning: A review and new perspectives, 2014. URLhttps://arxiv.org/abs/1206.5538

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  21. [21]

    Emerging Properties in Self-Supervised Vision Transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerging properties in self-supervised vision transformers, 2021. URL https://arxiv.org/abs/2104.14294

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  22. [22]

    Concept bottleneck models, 2020

    Pang Wei Koh, Thao Nguyen, Y ew Siang Tang, Stephen Mussmann, Emma Pierson, Been Kim, and Percy Liang. Concept bottleneck models, 2020. URLhttps://arxiv.org/abs/ 2007.04612

    work page arXiv 2020

  23. [23]

    Explaining classifiers with causal concept effect (cace), 2020

    Y ash Goyal, Amir Feder, Uri Shalit, and Been Kim. Explaining classifiers with causal concept effect (cace), 2020. URLhttps://arxiv.org/abs/1907.07165

    work page arXiv 2020

  24. [24]

    Towards automatic concept- based explanations, 2019

    Amirata Ghorbani, James Wexler, James Zou, and Been Kim. Towards automatic concept- based explanations, 2019. URLhttps://arxiv.org/abs/1902.03129. 23

    work page arXiv 2019

  25. [25]

    Got- way, and Jianming Liang

    Fatemeh Haghighi, Mohammad Reza Hosseinzadeh Taher, Zongwei Zhou, Michael B. Got- way, and Jianming Liang. Transferable visual words: Exploiting the semantics of anatomical patterns for self-supervised learning.IEEE Transactions on Medical Imaging, 40(10):2857– 2868, 2021. doi:10.1109/TMI.2021.3060634

    work page doi:10.1109/tmi.2021.3060634 2021

  26. [26]

    HGCLIP: Exploring vision-language models with graph representations for hierarchical understanding

    Peng Xia, Xingtong Yu, Ming Hu, Lie Ju, Zhiyong Wang, Peibo Duan, and Zongyuan Ge. HGCLIP: Exploring vision-language models with graph representations for hierarchical understanding. In Owen Rambow, Leo Wanner, Marianna Apidianaki, Hend Al-Khalifa, Barbara Di Eugenio, and Steven Schockaert, editors,Proceedings of the 31st Interna- tional Conference on Com...

    work page

  27. [27]

    URLhttps://aclanthology.org/2025

    Association for Computational Linguistics. URLhttps://aclanthology.org/2025. coling-main.19/

    work page 2025

  28. [28]

    Radalign: Advancing radiology report generation with vision-language concept alignment, 2025

    Difei Gu, Yunhe Gao, Y ang Zhou, Mu Zhou, and Dimitris Metaxas. Radalign: Advancing radiology report generation with vision-language concept alignment, 2025. URLhttps: //arxiv.org/abs/2501.07525

    work page arXiv 2025

  29. [29]

    The cancer genome atlas pan-cancer analysis project.Nature Genetics, 45(10):1113–1120, 2013

    John N Weinstein et al. The cancer genome atlas pan-cancer analysis project.Nature Genetics, 45(10):1113–1120, 2013

    work page 2013

  30. [30]

    The cancer proteome atlas (cptac): a community resource for proteogenomic cancer research.Journal of Proteome Research, 14(6):2707–2713, 2015

    Nathan J Edwards et al. The cancer proteome atlas (cptac): a community resource for proteogenomic cancer research.Journal of Proteome Research, 14(6):2707–2713, 2015

    work page 2015

  31. [31]

    Integrated genomic and molecular charac- terization of cervical cancer.Nature, 543:378–384, 2017

    The Cancer Genome Atlas Research Network. Integrated genomic and molecular charac- terization of cervical cancer.Nature, 543:378–384, 2017. doi:10.1038/nature21386

    work page doi:10.1038/nature21386 2017

  32. [32]

    Gutman, Jill S

    Pooya Mobadersany, Sahand Y ousefi, Mohamed Amgad, David A. Gutman, Jill S. Barnholtz-Sloan, Jorge E. Vel ´azquez Vega, Daniel J. Brat, and Lee A. D. Cooper. Pre- dicting cancer outcomes from histology and genomics using convolutional networks.Pro- ceedings of the National Academy of Sciences, 115(13):E2970–E2979, Mar 2018. doi: 10.1073/pnas.1717139115

    work page doi:10.1073/pnas.1717139115 2018

  33. [33]

    Measuring domain shift for deep learning in histopathology.IEEE Journal of Biomedical and Health Informatics, 25(2):325–336, 2021

    Kristofer Stacke, Gabriel Eilertsen, Jonas Unger, and Claes Lundstr ¨om. Measuring domain shift for deep learning in histopathology.IEEE Journal of Biomedical and Health Informatics, 25(2):325–336, 2021. doi:10.1109/JBHI.2020.3032060

    work page doi:10.1109/jbhi.2020.3032060 2021

  34. [34]

    Zarella, Deborah Bowman, Famke Aeffner, Navid Farahani, Andrew Xthona, Syed F

    Mark D. Zarella, Deborah Bowman, Famke Aeffner, Navid Farahani, Andrew Xthona, Syed F . Absar, Anil Parwani, Marilyn Bui, and Douglas J. Hartman. A practical guide to whole slide imaging: A white paper from the digital pathology association.Archives of Pathology & Laboratory Medicine, 143(2):222–234, 2019. doi:10.5858/arpa.2018-0343-RA

    work page doi:10.5858/arpa.2018-0343-ra 2019

  35. [35]

    Self supervised contrastive learning for digital histopathology.Machine Learning with Applications, 7:100198, 2022

    Ozan Ciga, Tony Xu, and Anne Louise Martel. Self supervised contrastive learning for digital histopathology.Machine Learning with Applications, 7:100198, 2022. ISSN 2666-

    work page 2022

  36. [36]

    URLhttps://www.sciencedirect

    doi:https://doi.org/10.1016/j.mlwa.2021.100198. URLhttps://www.sciencedirect. com/science/article/pii/S2666827021000992

    work page doi:10.1016/j.mlwa.2021.100198 2021

  37. [37]

    Tissue concepts v2: A supervised foundation model for whole slide images, 2025

    Till Nicke, Daniela Schacherer, Jan Raphael Sch ¨afer, Natalia Artysh, Antje Prasse, Andr ´e Homeyer, Andrea Schenk, Henning H ¨ofener, and Johannes Lotz. Tissue concepts v2: A supervised foundation model for whole slide images, 2025. URLhttps://arxiv.org/abs/ 2507.05742

    work page arXiv 2025

  38. [38]

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

    Richard J Chen, Tong Ding, Ming Y Lu, Drew FK Williamson, Guillaume Jaume, Bowen Chen, Andrew Zhang, Daniel Shao, Andrew H Song, Muhammad Shaban, et al. Towards a general-purpose foundation model for computational pathology.Nature Medicine, 2024. 24

    work page 2024

  39. [39]

    Lu, Bowen Chen, Drew F

    Ming Y . Lu, Bowen Chen, Drew F . K. Williamson, Richard J. Chen, Isaac Liang, Tian Ding, Guillaume Jaume, Igor Odintsov, Long Phi Le, Georg Gerber, Anil V. Parwani, Ann Zhang, and Faisal Mahmood. A visual-language foundation model for computational pathology. Nature Medicine, 30(3):863–874, March 2024. doi:10.1038/s41591-024-02856-4

    work page doi:10.1038/s41591-024-02856-4 2024

  40. [40]

    Wright, Ari Robicsek, Brian Piening, Carlo Bifulco, Sheng Wang, and Hoifung Poon

    Hanwen Xu, Naoto Usuyama, Jaspreet Bagga, Sheng Zhang, Rajesh Rao, Tristan Nau- mann, Cliff Wong, Zelalem Gero, Javier Gonz ´alez, Yu Gu, Y anbo Xu, Mu Wei, Wenhui Wang, Shuming Ma, Furu Wei, Jianwei Y ang, Chunyuan Li, Jianfeng Gao, Jaylen Rose- mon, Tucker Bower, Soohee Lee, Roshanthi Weerasinghe, Bill J. Wright, Ari Robicsek, Brian Piening, Carlo Biful...

    work page 2024

  41. [41]

    Emre Celebi, Hassan A

    M. Emre Celebi, Hassan A. Kingravi, and Patricio A. Vela. A comparative study of efficient initialization methods for the k-means clustering algorithm.Expert Systems with Applica- tions, 40(1):200–210, January 2013. ISSN 0957-4174. doi:10.1016/j.eswa.2012.07.021. URLhttp://dx.doi.org/10.1016/j.eswa.2012.07.021

    work page doi:10.1016/j.eswa.2012.07.021 2013

  42. [42]

    Comprehensive k-means clustering.Journal of Computer and Communica- tions, 12(3), March 26 2024

    Ethan Xiao. Comprehensive k-means clustering.Journal of Computer and Communica- tions, 12(3), March 26 2024

    work page 2024

  43. [43]

    S. Lloyd. Least squares quantization in pcm.IEEE Transactions on Information Theory, 28 (2):129–137, 1982. doi:10.1109/TIT.1982.1056489

    work page doi:10.1109/tit.1982.1056489 1982

  44. [44]

    Probabilistic conceptual explainers: Trustwor- thy conceptual explanations for vision foundation models

    Hengyi Wang, Shiwei Tan, and Hao Wang. Probabilistic conceptual explainers: Trustwor- thy conceptual explanations for vision foundation models. InInternational Conference on Machine Learning, 2024

    work page 2024

  45. [45]

    Thorndike

    Robert L. Thorndike. Who belongs in the family?Psychometrika, 18(4):267–276, 1953. 25

    work page 1953