pith. sign in

arxiv: 2405.04211 · v3 · submitted 2024-05-07 · 💻 cs.CV

Leveraging Medical Foundation Model Features in Graph Neural Network-Based Retrieval of Breast Histopathology Images

Nematollah Saeidi , Hossein Karshenas , Bijan Shoushtarian , Sepideh Hatamikia , Ramona Woitek , Amirreza Mahbod This is my paper

Pith reviewed 2026-05-24 01:07 UTC · model grok-4.3

classification 💻 cs.CV
keywords breast cancerhistopathologyimage retrievalfoundation modelsgraph neural networksvariational autoencodercomputational pathology
0
0 comments X

The pith

Foundation model features in a graph autoencoder improve breast histopathology image retrieval over CNN baselines.

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

The paper proposes using features from medical foundation models inside an attention-based adversarially regularized variational graph autoencoder to retrieve similar breast cancer histology images. On the BreakHis and BACH datasets the foundation-model versions reach higher mean average precision and mean maximum visibility scores than versions that rely on pre-trained convolutional neural network features, with the UNI pathology model performing best. A sympathetic reader would care because more accurate automated retrieval could help pathologists locate matching tissue patterns and shorten diagnostic time. The work reports concrete gains of up to 7.7 percent mAP and 15.5 percent mMV when foundation features replace CNN features.

Core claim

The central claim is that an attention-based adversarially regularized variational graph autoencoder trained on features from foundation models, especially the self-supervised UNI model, produces higher retrieval accuracy than the same architecture trained on features from pre-trained convolutional neural networks, reaching average mAP/mMV of 96.7 percent/91.5 percent on BreakHis and 97.6 percent/94.2 percent on BACH.

What carries the argument

Attention-based adversarially regularized variational graph autoencoder that ingests foundation-model embeddings to encode tissue variability for image retrieval.

Load-bearing premise

The gains measured on two public datasets will continue to appear on new clinical images from different scanners or hospitals.

What would settle it

Evaluation of the same model on an independent, previously unseen breast histopathology dataset where mAP and mMV fall below the CNN-feature baseline.

Figures

Figures reproduced from arXiv: 2405.04211 by Amirreza Mahbod, Bijan Shoushtarian, Hossein Karshenas, Nematollah Saeidi, Ramona Woitek, Sepideh Hatamikia.

Figure 1
Figure 1. Figure 1: The proposed general workflow of the breast histological image retrieval model [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Example images from the BreakHis (first four images) and the BACH (last [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The framework of graph construction. K-ANN: K approximate nearest neigh￾bors, FLANN: fast library for approximate nearest neighbors is more time-efficient than the k-NN algorithm and beneficial, especially in high-dimensional feature spaces, where the complexity of high dimensions slows down exact nearest neighbor searches. While there are various imple￾mentations of ANN algorithms, the ANN benchmark [52] … view at source ↗
Figure 4
Figure 4. Figure 4: The Architecture of attention-based adversarially regularized variational graph [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Five sample queries from the BreakHis and BACH datasets and their top similar [PITH_FULL_IMAGE:figures/full_fig_p018_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of embedding values for four image pairs. Similar pairs refer to [PITH_FULL_IMAGE:figures/full_fig_p019_6.png] view at source ↗
read the original abstract

Breast cancer is the most common cancer type in women worldwide. Early detection and appropriate treatment can significantly reduce its impact. While histopathology examinations play a vital role in rapid and accurate diagnosis, they often require experienced medical experts for proper recognition and cancer grading. Automated image retrieval systems have the potential to assist pathologists in identifying cancerous tissues, thereby accelerating the diagnostic process. Nevertheless, proposing an accurate image retrieval model is challenging due to considerable variability among the tissue and cell patterns in histological images. In this work, we leverage the features from foundation models in a novel attention-based adversarially regularized variational graph autoencoder model for breast histological image retrieval. Our results confirm the superior performance of models trained with foundation model features compared to those using pre-trained convolutional neural networks (up to 7.7% and 15.5% for mAP and mMV, respectively), with the pre-trained general-purpose self-supervised model for computational pathology (UNI) delivering the best overall performance. By evaluating two publicly available histology image datasets of breast cancer, our top-performing model, trained with UNI features, achieved average mAP/mMV scores of 96.7%/91.5% and 97.6%/94.2% for the BreakHis and BACH datasets, respectively. Our proposed retrieval model has the potential to be used in clinical settings to enhance diagnostic performance and ultimately benefit patients.

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

2 major / 1 minor

Summary. The paper introduces an attention-based adversarially regularized variational graph autoencoder for content-based retrieval of breast histopathology images. It extracts features from medical foundation models (with UNI performing best) and reports that these yield higher retrieval accuracy than pre-trained CNN features, with gains of up to 7.7% mAP and 15.5% mMV; on BreakHis the best model reaches 96.7%/91.5% and on BACH 97.6%/94.2%.

Significance. If the empirical gains prove robust under proper statistical controls and external validation, the work would demonstrate a practical way to combine self-supervised pathology foundation models with graph autoencoders for improved retrieval, which could support diagnostic assistance tools in computational pathology.

major comments (2)
  1. [Results] Results section: performance is reported solely as point estimates (e.g., 96.7% mAP, 91.5% mMV on BreakHis) with no error bars, standard deviations across random seeds, or statistical significance tests, so it is impossible to determine whether the claimed 7.7% and 15.5% margins over CNN baselines are stable or could arise from training variability.
  2. [Experiments] Experimental protocol (Methods/Experiments): the manuscript supplies no description of the train/validation/test split strategy (patient-level vs. image-level), number of independent runs, hyperparameter selection procedure, or external-site validation, leaving open the possibility that the reported superiority of UNI features is tied to the specific public datasets and splits used.
minor comments (1)
  1. [Abstract] The abbreviation mMV is used without an explicit definition in the abstract or early sections; a one-sentence expansion would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on statistical robustness and experimental transparency. We address each major comment below and will revise the manuscript to incorporate the suggested improvements.

read point-by-point responses

  1. Referee: [Results] Results section: performance is reported solely as point estimates (e.g., 96.7% mAP, 91.5% mMV on BreakHis) with no error bars, standard deviations across random seeds, or statistical significance tests, so it is impossible to determine whether the claimed 7.7% and 15.5% margins over CNN baselines are stable or could arise from training variability.

    Authors: We agree that point estimates alone are insufficient. In the revised manuscript we will report mean performance and standard deviation across five independent runs with different random seeds, add error bars to all tables and figures, and include statistical significance tests (paired t-tests or Wilcoxon signed-rank tests) against the CNN baselines to confirm the reported margins are stable. revision: yes

  2. Referee: [Experiments] Experimental protocol (Methods/Experiments): the manuscript supplies no description of the train/validation/test split strategy (patient-level vs. image-level), number of independent runs, hyperparameter selection procedure, or external-site validation, leaving open the possibility that the reported superiority of UNI features is tied to the specific public datasets and splits used.

    Authors: We will expand the Methods and Experiments sections with a complete protocol description. Patient-level splits were used (70/15/15 train/val/test) to avoid leakage; hyperparameters were selected via grid search on the validation set; five independent runs were performed. Exact split indices and seeds will be released. External-site validation was outside the current scope and will be noted as a limitation, but the public datasets enable full reproducibility. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical evaluation on public datasets

full rationale

The paper proposes an attention-based adversarially regularized variational graph autoencoder that ingests features from foundation models (e.g., UNI) or pre-trained CNNs and reports retrieval metrics (mAP, mMV) on the BreakHis and BACH datasets. No equation, prediction, or uniqueness claim reduces by construction to a fitted parameter, self-defined quantity, or self-citation chain. All load-bearing statements are direct experimental outcomes on fixed public benchmarks; the derivation chain is therefore self-contained and does not exhibit any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The performance claim depends on the transferability of UNI features to the retrieval task and on the effectiveness of the chosen graph-autoencoder architecture; both are treated as domain assumptions rather than derived quantities.

free parameters (1)
  • graph autoencoder hyperparameters
    Architecture depth, attention heads, adversarial regularization strength, and latent dimension are tuned to produce the reported mAP and mMV values on the two datasets.
axioms (1)
  • domain assumption Features extracted from the pre-trained UNI foundation model are directly suitable as node attributes for the graph autoencoder without further adaptation or domain-specific fine-tuning.
    The abstract presents UNI features as the best-performing input but does not derive or justify their transferability beyond the empirical outcome.

pith-pipeline@v0.9.0 · 5807 in / 1403 out tokens · 27007 ms · 2026-05-24T01:07:09.014353+00:00 · methodology

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

63 extracted references · 63 canonical work pages · 4 internal anchors

  1. [1]

    Rashmi, K

    R. Rashmi, K. Prasad, C. B. K. Udupa, Breast histopathological image analysis using image processing techniques for diagnostic puposes: A methodological review., Journal of medical systems 46 (1) (2021) 7.doi: https://doi.org/10.1007/s10916-021-01786-9

    work page doi:10.1007/s10916-021-01786-9 2021

  2. [2]

    Arnold, E

    M. Arnold, E. Morgan, H. Rumgay, A. Mafra, D. Singh, M. Laver- sanne, J. Vignat, J. R. Gralow, F. Cardoso, S. Siesling, I. Soerjo- mataram, Current and future burden of breast cancer: Global statis- tics for 2020 and 2040, The Breast 66 (2022) 15–23. doi:https: //doi.org/10.1016/j.breast.2022.08.010

    work page doi:10.1016/j.breast.2022.08.010 2020

  3. [3]

    A. E. Minarno, K. M. Ghufron, T. S. Sabrila, L. Husniah, F. D. S. Sumadi, CNN based autoencoder application in breast cancer im- age retrieval, in: International Seminar on Intelligent Technology and Its Applications, 2021, pp. 29–34. doi:https://doi.org/10.1109/ ISITIA52817.2021.9502205

    work page arXiv 2021

  4. [4]

    Burstein, G

    H. Burstein, G. Curigliano, S. Loibl, P. Dubsky, M. Gnant, P. Poort- mans, M. Colleoni, C. Denkert, M. Piccart-Gebhart, M. Regan, H.-J. Senn, E. Winer, B. Thurlimann, Estimating the benefits of therapy for early-stage breast cancer: the st. gallen international consensus guide- lines for the primary therapy of early breast cancer 2019, Annals of Oncology ...

    work page 2019

  5. [5]

    Tabatabaei, A

    Z. Tabatabaei, A. Colomer, K. Engan, J. Oliver, V. Naranjo, Resid- ual block convolutional auto encoder in content-based medical image retrieval, in: IEEE 14th Image, Video, and Multidimensional Signal Processing Workshop, 2022, pp. 1–5. doi:https://doi.org/10.1109/ IVMSP54334.2022.9816325

    work page arXiv 2022

  6. [6]

    Fuster, F

    S. Fuster, F. Khoraminia, U. Kiraz, N. Kanwal, V. Kvikstad, T. Eftestøl, T. C. Zuiverloon, E. A. Janssen, K. Engan, Invasive cancerous area de- tection in non-muscle invasive bladder cancer whole slide images, in: 21 IEEE 14th Image, Video, and Multidimensional Signal Processing Work- shop, 2022, pp. 1–5. doi:https://doi.org/10.1109/IVMSP54334. 2022.9816352

    work page doi:10.1109/ivmsp54334 2022

  7. [7]

    Agrawal, A

    D. Agrawal, A. Agarwal, D. K. Sharma, Content-based image retrieval (cbir): A review, in: P. K. Singh, Y. Singh, J. K. Chhabra, Z. Ill´ es, C. Verma (Eds.), Recent Innovations in Computing, Springer Singa- pore, Singapore, 2022, pp. 439–452. doi:https://doi.org/10.1007/ 978-981-16-8892-8_33

    work page 2022

  8. [8]

    Y. A. Malkov, D. A. Yashunin, Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs, IEEE Transactions on Pattern Analysis and Machine Intelligence 42 (4) (2020) 824–836. doi:https://doi.org/10.1109/TPAMI.2018.2889473

    work page doi:10.1109/tpami.2018.2889473 2020

  9. [9]

    L. Z, Z. X, M. H, Z. S, Large-scale retrieval for medical image analytics: A comprehensive review, Medical image analysis 43 (2018) 66–84. doi: https://doi.org/10.1016/J.MEDIA.2017.09.007

    work page doi:10.1016/j.media.2017.09.007 2018

  10. [10]

    Silva-Rodr´ ıguez, A

    J. Silva-Rodr´ ıguez, A. Colomer, M. A. Sales, R. Molina, V. Naranjo, Going deeper through the Gleason scoring scale: An automatic end-to- end system for histology prostate grading and cribriform pattern de- tection, Computer Methods and Programs in Biomedicine 195 (2020). doi:https://doi.org/10.1016/j.cmpb.2020.105637

    work page doi:10.1016/j.cmpb.2020.105637 2020

  11. [11]

    Hegde, J

    N. Hegde, J. D. Hipp, Y. Liu, M. Emmert-Buck, E. Reif, D. Smilkov, M. Terry, C. J. Cai, M. B. Amin, C. H. Mermel, P. Q. Nelson, L. H. Peng, G. S. Corrado, M. C. Stumpe, Similar image search for histopathology: SMILY, npj Digital Medicine 2 (1) (2019). doi:https://doi.org/10. 1038/s41746-019-0131-z

    work page 2019

  12. [12]

    Yamashita, M

    R. Yamashita, M. Nishio, R. K. G. Do, K. Togashi, Convolutional neural networks: an overview and application in radiology (2018). doi:https: //doi.org/10.1007/s13244-018-0639-9

    work page doi:10.1007/s13244-018-0639-9 2018

  13. [13]

    Xiong, J

    Z. Xiong, J. Cai, Multi-scale Graph Convolutional Networks with Self- Attention, arXiv preprint arXiv:2112.03262 (2021)

    work page arXiv 2021

  14. [14]

    S. Pan, R. Hu, G. Long, J. Jiang, L. Yao, C. Zhang, Adversarially regularized graph autoencoder for graph embedding, in: IJCAI Interna- 22 tional Joint Conference on Artificial Intelligence, Vol. 2018-July, 2018. doi:https://doi.org/10.24963/ijcai.2018/362

    work page doi:10.24963/ijcai.2018/362 2018

  15. [15]

    Zheng, J

    M. Zheng, J. Xu, Y. Shen, C. Tian, J. Li, L. Fei, M. Zong, X. Liu, Attention-based CNNs for Image Classification: A Survey, in: Journal of Physics: Conference Series, Vol. 2171, 2022. doi:https://doi.org/ 10.1088/1742-6596/2171/1/012068

    work page doi:10.1088/1742-6596/2171/1/012068 2022

  16. [16]

    H. Xia, S. Shao, C. Hu, R. Zhang, T. Qiu, F. Xiao, Robust clustering model based on attention mechanism and graph convolutional network, IEEE Transactions on Knowledge and Data Engineering 35 (5) (2023) 5203–5215. doi:https://doi.org/10.1109/TKDE.2022.3150300

    work page doi:10.1109/tkde.2022.3150300 2023

  17. [17]

    Z. Weng, W. Zhang, W. Dou, Adversarial Attention-Based Variational Graph Autoencoder, IEEE Access 8 (2020). doi:https://doi.org/10. 1109/ACCESS.2020.3018033

    work page arXiv 2020

  18. [18]

    J. Xiao, Q. Dai, X. Xie, J. Lam, K. W. Kwok, Adversarially reg- ularized graph attention networks for inductive learning on partially labeled graphs, Knowledge-Based Systems 268 (2023). doi:https: //doi.org/10.1016/j.knosys.2023.110456

    work page doi:10.1016/j.knosys.2023.110456 2023

  19. [19]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, L. Fei-Fei, ImageNet: A large-scale hierarchical image database, in: IEEE Conference on Com- puter Vision and Pattern Recognition, 2009, pp. 248–255. doi:https: //doi.org/10.1109/CVPR.2009.5206848

    work page doi:10.1109/cvpr.2009.5206848 2009

  20. [20]

    T. N. Kipf, M. Welling, Variational graph auto-encoders, arXiv preprint arXiv:1611.07308 (2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  21. [21]

    Denner, D

    S. Denner, D. Zimmerer, D. Bounias, M. Bujotzek, S. Xiao, L. Kausch, P. Schader, T. Penzkofer, P. F. J¨ ager, K. Maier-Hein, Leveraging foun- dation models for content-based medical image retrieval in radiology, arXiv preprint arXiv:2403.06567 (2024)

    work page arXiv 2024

  22. [22]

    BiomedCLIP: a multimodal biomedical foundation model pretrained from fifteen million scientific image-text pairs

    S. Zhang, Y. Xu, N. Usuyama, H. Xu, J. Bagga, R. Tinn, S. Preston, R. Rao, M. Wei, N. Valluri, et al., BiomedCLIP: a multimodal biomedi- cal foundation model pretrained from fifteen million scientific image-text pairs, arXiv preprint arXiv:2303.00915 (2023). 23

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  23. [23]

    R. J. Chen, T. Ding, M. Y. Lu, D. F. K. Williamson, G. Jaume, A. H. Song, B. Chen, A. Zhang, D. Shao, M. Shaban, M. Williams, L. Old- enburg, L. L. Weishaupt, J. J. Wang, A. Vaidya, L. P. Le, G. Gerber, S. Sahai, W. Williams, F. Mahmood, Towards a general-purpose founda- tion model for computational pathology, Nature Medicine 30 (3) (2024) 850–862. doi:ht...

    work page doi:10.1038/s41591-024-02857-3 2024

  24. [24]

    X. Wang, Y. Du, S. Yang, J. Zhang, M. Wang, J. Zhang, W. Yang, J. Huang, X. Han, RetCCL: Clustering-guided contrastive learning for whole-slide image retrieval, Medical Image Analysis 83 (2023). doi: https://doi.org/10.1016/j.media.2022.102645

    work page doi:10.1016/j.media.2022.102645 2023

  25. [25]

    F. A. Spanhol, L. S. Oliveira, C. Petitjean, L. Heutte, A Dataset for Breast Cancer Histopathological Image Classification, IEEE Transac- tions on Biomedical Engineering 63 (7) (2016). doi:https://doi.org/ 10.1109/TBME.2015.2496264

    work page doi:10.1109/tbme.2015.2496264 2016

  26. [26]

    Aresta, T

    G. Aresta, T. Ara´ ujo, S. Kwok, S. S. Chennamsetty, M. Safwan, V. Alex, B. Marami, M. Prastawa, M. Chan, M. Donovan, G. Fernan- dez, J. Zeineh, M. Kohl, C. Walz, F. Ludwig, S. Braunewell, M. Baust, Q. D. Vu, M. N. N. To, E. Kim, J. T. Kwak, S. Galal, V. Sanchez-Freire, N. Brancati, M. Frucci, D. Riccio, Y. Wang, L. Sun, K. Ma, J. Fang, I. Kone, L. Boulma...

    work page doi:10.1016/j.media 2019

  27. [27]

    N. T. Singh, C. Kaur, A. Chaudhary, S. Goyal, Preprocessing of med- ical images using deep learning: A comprehensive review, in: Interna- tional Conference on Augmented Intelligence and Sustainable Systems, 2023, pp. 521–527. doi:https://doi.org/10.1109/ICAISS58487. 2023.10250462

    work page doi:10.1109/icaiss58487 2023

  28. [28]

    Murcia-G´ omez, I

    D. Murcia-G´ omez, I. Rojas-Valenzuela, O. Valenzuela, Impact of im- age preprocessing methods and deep learning models for classifying histopathological breast cancer images, Applied Sciences 12 (22) (2022). doi:https://doi.org/10.3390/app122211375

    work page doi:10.3390/app122211375 2022

  29. [29]

    M. Tan, Q. Le, EfficientNetV2: Smaller models and faster training, in: M. Meila, T. Zhang (Eds.), Proceedings of the 38th International 24 Conference on Machine Learning, Vol. 139 of Proceedings of Machine Learning Research, PMLR, 2021, pp. 10096–10106

    work page 2021

  30. [30]

    Densely connected convolutional networks,

    G. Huang, Z. Liu, L. Van Der Maaten, K. Q. Weinberger, Densely connected convolutional networks, in: IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 2261–2269. doi:https: //doi.org/10.1109/CVPR.2017.243

    work page doi:10.1109/cvpr.2017.243 2017

  31. [31]

    C. C. Ukwuoma, M. A. Hossain, J. K. Jackson, G. U. Nneji, H. N. Monday, Z. Qin, Multi-Classification of Breast Cancer Lesions in Histopathological Images Using DEEP Pachi: Multiple Self-Attention Head, Diagnostics 12 (5) (2022). doi:https://doi.org/10.3390/ diagnostics12051152

    work page 2022

  32. [32]

    Mahbod, G

    A. Mahbod, G. Schaefer, R. Ecker, I. Ellinger, Pollen grain micro- scopic image classification using an ensemble of fine-tuned deep convolu- tional neural networks, in: International Conference on Pattern Recog- nition, Springer, 2021, pp. 344–356. doi:https://doi.org/10.1007/ 978-3-030-68763-2_26

    work page 2021

  33. [33]

    D. A. Suju, H. Jose, FLANN: Fast approximate nearest neighbour search algorithm for elucidating human-wildlife conflicts in forest areas, in: 2017 4th International Conference on Signal Processing, Communica- tion and Networking, ICSCN 2017, 2017. doi:https://doi.org/10. 1109/ICSCN.2017.8085676

    work page arXiv 2017

  34. [34]

    Kalra, H

    S. Kalra, H. R. Tizhoosh, C. Choi, S. Shah, P. Diamandis, C. J. Camp- bell, L. Pantanowitz, Yottixel – an image search engine for large archives of histopathology whole slide images, Medical Image Analysis 65 (2020) 101757. doi:https://doi.org/10.1016/J.MEDIA.2020.101757

    work page doi:10.1016/j.media.2020.101757 2020

  35. [35]

    C. Chen, M. Y. Lu, D. F. Williamson, T. Y. Chen, A. J. Schaumberg, F. Mahmood, Fast and scalable search of whole-slide images via self- supervised deep learning, Nature Biomedical Engineering 6 (12) (2022). doi:https://doi.org/10.1038/s41551-022-00929-8

    work page doi:10.1038/s41551-022-00929-8 2022

  36. [36]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, I. Sutskever, Learning transferable visual models from natural language supervision, in: M. Meila, T. Zhang (Eds.), Proceedings of the 38th International 25 Conference on Machine Learning, Vol. 139 of roceedings of Machine Learning Res...

    work page 2021

  37. [37]

    Walk in the cloud: Learning curves for point clouds shape analysis, pp

    M. Caron, H. Touvron, I. Misra, H. Jegou, J. Mairal, P. Bojanowski, A. Joulin, Emerging properties in self-supervised vision transformers, in: IEEE/CVF International Conference on Computer Vision, 2021, pp. 9630–9640. doi:https://doi.org/10.1109/ICCV48922.2021.00951

    work page doi:10.1109/iccv48922.2021.00951 2021

  38. [38]

    S. Yan, Z. Yu, C. Primiero, C. Vico-Alonso, Z. Wang, L. Yang, P. Tschandl, M. Hu, G. Tan, V. Tang, et al., A general- purpose multimodal foundation model for dermatology, arXiv preprint arXiv:2410.15038 (2024)

    work page arXiv 2024

  39. [39]

    Anand, S

    D. Anand, S. Gadiya, A. Sethi, Histographs: graphs in histopathol- ogy, in: J. E. Tomaszewski, A. D. Ward (Eds.), Medical Imaging 2020: Digital Pathology, Vol. 11320, SPIE, 2020, p. 113200O. doi:https: //doi.org/10.1117/12.2550114

    work page doi:10.1117/12.2550114 2020

  40. [40]

    Ahmedt-Aristizabal, M

    D. Ahmedt-Aristizabal, M. A. Armin, S. Denman, C. Fookes, L. Petersson, A survey on graph-based deep learning for computa- tional histopathology, Computerized Medical Imaging and Graphics 95 (2022) 102027. doi:https://doi.org/10.1016/j.compmedimag. 2021.102027

    work page doi:10.1016/j.compmedimag 2022

  41. [41]

    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 classifi- cation of nuclei in multi-tissue histology images, Medical Image Analy- sis 58 (2019) 101563. doi:https://doi.org/10.1016/j.media.2019. 101563

    work page doi:10.1016/j.media.2019 2019

  42. [42]

    Mahbod, G

    A. Mahbod, G. Schaefer, G. Dorffner, S. Hatamikia, R. Ecker, I. Ellinger, A dual decoder u-net-based model for nuclei instance segmentation in hematoxylin and eosin-stained histological images, Frontiers in Medicine 9 (2022). doi:https://doi.org/10.3389/fmed.2022.978146

    work page doi:10.3389/fmed.2022.978146 2022

  43. [43]

    Ahmedt-Aristizabal, M

    D. Ahmedt-Aristizabal, M. A. Armin, S. Denman, C. Fookes, L. Pe- tersson, Graph-based deep learning for medical diagnosis and analy- sis: Past, present and future, Sensors 21 (14) (2021). doi:https: //doi.org/10.3390/s21144758. 26

    work page doi:10.3390/s21144758 2021

  44. [44]

    Zheng, Z

    Y. Zheng, Z. Jiang, H. Zhang, F. Xie, Y. Ma, H. Shi, Y. Zhao, Histopathological Whole Slide Image Analysis Using Context-Based CBIR, IEEE Transactions on Medical Imaging 37 (7) (2018). doi: https://doi.org/10.1109/TMI.2018.2796130

    work page doi:10.1109/tmi.2018.2796130 2018

  45. [45]

    Zheng, B

    Y. Zheng, B. Jiang, J. Shi, H. Zhang, F. Xie, Encoding histopathological wsis using gnn for scalable diagnostically relevant regions retrieval, in: D. Shen, T. Liu, T. M. Peters, L. H. Staib, C. Essert, S. Zhou, P.-T. Yap, A. Khan (Eds.), Medical Image Computing and Computer Assisted Intervention, Springer International Publishing, Cham, 2019, pp. 550–

    work page 2019

  46. [46]

    doi:https://doi.org/10.1007/978-3-030-32239-7_61

    work page doi:10.1007/978-3-030-32239-7_61

  47. [47]

    DINOv2: Learning Robust Visual Features without Supervision

    M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec, V. Khali- dov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby, et al., DINOv2: Learning robust visual features without supervision, arXiv preprint arXiv:2304.07193 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  48. [48]

    Simonyan, A

    K. Simonyan, A. Zisserman, Very deep convolutional networks for large- scale image recognition, Computational and Biological Learning Society, 2015, pp. 1–14

    work page 2015

  49. [49]

    Sandler, A

    M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, L.-C. Chen, Mo- bileNetV2: Inverted residuals and linear bottlenecks, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,

    work page

  50. [50]

    doi:https://doi.org/10.1109/CVPR.2018.00474

    work page doi:10.1109/cvpr.2018.00474 2018

  51. [51]

    B. Zoph, V. Vasudevan, J. Shlens, Q. V. Le, Learning transferable ar- chitectures for scalable image recognition, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018, pp. 8697–8710. doi:https://doi.org/10.1109/CVPR.2018.00907

    work page doi:10.1109/cvpr.2018.00907 2018

  52. [52]

    M. Lin, K. Wen, X. Zhu, H. Zhao, X. Sun, Graph Autoencoder with Pre- serving Node Attribute Similarity, Entropy 25 (4) (2023). doi:https: //doi.org/10.3390/e25040567

    work page doi:10.3390/e25040567 2023

  53. [53]

    Z. Gao, Z. Lu, J. Wang, S. Ying, J. Shi, A Convolutional Neural Net- work and Graph Convolutional Network Based Framework for Classifi- cation of Breast Histopathological Images, IEEE Journal of Biomedical and Health Informatics 26 (7) (2022). doi:https://doi.org/10.1109/ JBHI.2022.3153671. 27

    work page arXiv 2022

  54. [54]

    Aum¨ uller, E

    M. Aum¨ uller, E. Bernhardsson, A. Faithfull, Ann-benchmarks: A bench- marking tool for approximate nearest neighbor algorithms, in: Similarity Search and Applications: 10th International Conference, SISAP 2017, Munich, Germany, October 4-6, 2017, Proceedings 10, Springer, 2017, pp. 34–49

    work page 2017

  55. [55]

    Graph Attention Networks

    P. Veliˇ ckovi´ c, G. Cucurull, A. Casanova, A. Romero, P. Lio, Y. Bengio, Graph attention networks, arXiv preprint arXiv:1710.10903 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  56. [56]

    Masset, R

    Y. Zheng, Z. Jiang, J. Shi, F. Xie, H. Zhang, W. Luo, D. Hu, S. Sun, Z. Jiang, C. Xue, Encoding histopathology whole slide images with location-aware graphs for diagnostically relevant regions retrieval, Med- ical Image Analysis 76 (2022). doi:https://doi.org/10.1016/j. media.2021.102308

    work page doi:10.1016/j 2022

  57. [57]

    Parcham, M

    E. Parcham, M. Ilbeygi, M. Amini, CBCapsNet: A novel writer- independent offline signature verification model using a cnn-based ar- chitecture and capsule neural networks, Expert Systems with Appli- cations 185 (2021) 115649. doi:https://doi.org/10.1016/j.eswa. 2021.115649

    work page doi:10.1016/j.eswa 2021

  58. [58]

    S. Yun, M. Jeong, S. Yoo, S. Lee, S. S. Yi, R. Kim, J. Kang, H. J. Kim, Graph transformer networks: Learning meta-path graphs to improve gnns, Neural Networks 153 (2022) 104–119. doi:https://doi.org/ 10.1016/j.neunet.2022.05.026

    work page doi:10.1016/j.neunet.2022.05.026 2022

  59. [59]

    Johnson, M

    J. Johnson, M. Douze, H. J´ egou, Billion-scale similarity search with GPUs, IEEE Transactions on Big Data 7 (3) (2021) 535–547. doi: https://doi.org/10.1109/TBDATA.2019.2921572

    work page doi:10.1109/tbdata.2019.2921572 2021

  60. [60]

    Mahbod, N

    A. Mahbod, N. Saeidi, S. Hatamikia, R. Woitek, Evaluating pre- trained convolutional neural networks and foundation models as fea- ture extractors for content-based medical image retrieval, arXiv preprint arXiv:2409.09430 (2024)

    work page arXiv 2024

  61. [61]

    Moshkov, B

    N. Moshkov, B. Mathe, A. Kertesz-Farkas, R. Hollandi, P. Horvath, Test-time augmentation for deep learning-based cell segmentation on microscopy images, Scientific Reports 10 (1) (2020) 1–7. doi:https: //doi.org/10.1038/s41598-020-61808-3 . 28

    work page doi:10.1038/s41598-020-61808-3 2020

  62. [62]

    Mahbod, G

    A. Mahbod, G. Dorffner, I. Ellinger, R. Woitek, S. Hatamikia, Improv- ing generalization capability of deep learning-based nuclei instance seg- mentation by non-deterministic train time and deterministic test time stain normalization, Computational and Structural Biotechnology Jour- nal 23 (2024) 669–678. doi:https://doi.org/10.1016/j.csbj.2023. 12.042

    work page doi:10.1016/j.csbj.2023 2024

  63. [63]

    Bancher, A

    B. Bancher, A. Mahbod, I. Ellinger, R. Ecker, G. Dorffner, Improving mask r-cnn for nuclei instance segmentation in hematoxylin & eosin- stained histological images, in: MICCAI Workshop on Computational Pathology, Vol. 156, 2021, pp. 20–35. 29

    work page 2021