The DOI matches its Crossref record; 186 of the 186 references it cites were found in a registry or an index.
Towards a general-purpose foundation model for computational pathology
Open accessAbstract
Quantitative evaluation of tissue images is crucial for computational pathology (CPath) tasks, requiring the objective characterization of histopathological entities from whole-slide images (WSIs). The high resolution of WSIs and the variability of morphological features present significant challenges, complicating the large-scale annotation of data for high-performance applications. To address this challenge, current efforts have proposed the use of pretrained image encoders through transfer learning from natural image datasets or self-supervised learning on publicly available histopathology datasets, but have not been extensively developed and evaluated across diverse tissue types at scale. We introduce UNI, a general-purpose self-supervised model for pathology, pretrained using more than 100 million images from over 100,000 diagnostic H&E-stained WSIs (>77 TB of data) across 20 major tissue types. The model was evaluated on 34 representative CPath tasks of varying diagnostic difficulty. In addition to outperforming previous state-of-the-art models, we demonstrate new modeling capabilities in CPath such as resolution-agnostic tissue classification, slide classification using few-shot class prototypes, and disease subtyping generalization in classifying up to 108 cancer types in the OncoTree classification system. UNI advances unsupervised representation learning at scale in CPath in terms of both pretraining data and downstream evaluation, enabling data-efficient artificial intelligence models that can generalize and transfer to a wide range of diagnostically challenging tasks and clinical workflows in anatomic pathology. Pretrained using over 100,000 diagnostic histopathological slides across 20 major tissue types, a self-supervised model is shown to outperform existing baselines across various clinically relevant computational pathology tasks.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“By increasing the patch size from the widely used 256 × 256 pixels, we effectively reduce the sequence length by four without impacting the representation quality due to higher resolution patch input, leveraging the robustness of the patch-level foundation models in generalizing to higher resolutions9,10,87.”
From A multimodal whole-slide foundation model for pathology (2025) · cites this paper as [9]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“we present Virchow, the largest foundation model for computational pathology to date.”
From A foundation model for clinical-grade computational pathology and rare cancers detection (2024) · cites this paper as [41]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“To overcome these limitations, several research groups have alternatively proposed the use of weakly-supervised deep learning models, which can be trained to perform various downstream computational pathology tasks such as tumor subtyping and prognosis32–37 without any region or pixel-level annotations.”
From Predicting the tumor microenvironment composition and immunotherapy response in non-small cell lung cancer from digital histopathology images (2024) · cites this paper as [32]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“UNI [191], Prov-GigaPath [192], and Virchow [193] were pretrained solely on pathology image data to perform cancer classification, diagnosis, and prognosis.”
From Challenges in AI-driven Biomedical Multimodal Data Fusion and Analysis (2025) · cites this paper as [191]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Built on rich knowledge representations, these models have shown impressive performance across medical fields such as ophthalmology20, radiology21 and pathology22–25.”
From A Multimodal Vision Foundation Model for Clinical Dermatology (2025) · cites this paper as [23]
The cited source says what the claim attributes to it.
In the citing paper
“The vision module is based on the patch encoder CTransPath17 for the small and medium models, and UNI18 for the large model.”
From Generating dermatopathology reports from gigapixel whole slide images with HistoGPT (2025) · cites this paper as [18]
The cited source says what the claim attributes to it.
In the citing paper
“We found out that the UNI foundation model32, when paired with multi-task AB-MIL, achieved the best predictive performance for the various TME signature prediction tasks.”
From Predicting the tumor microenvironment composition and immunotherapy response in non-small cell lung cancer from digital histopathology images (2024) · cites this paper as [32]
In the citing paper
“In fact GMA is widely used to assess the performance of pathology foundation models23,25,26.”
From A clinical benchmark of public self-supervised pathology foundation models (2025) · cites this paper as [23]
In the citing paper
“Self-supervised learning has become a widely used approach for pretraining foundation models on thousands to tens of thousands of images without annotations or labels, marking an important paradigm shift in the field.20,21”
From Use of artificial intelligence for liver diseases: A survey from the EASL congress 2024 (2024) · cites this paper as [20]
In the citing paper
“An increasing number of published studies show the potential of foundation models20,24; however, they are currently underrepresented in the field of liver diseases.4”
From Use of artificial intelligence for liver diseases: A survey from the EASL congress 2024 (2024) · cites this paper as [20]
In the citing paper
“These works demonstrate that image features produced with self-supervised learning of pathology images outperform image features trained on natural images and that performance improves with scale.”
From A foundation model for clinical-grade computational pathology and rare cancers detection (2024) · cites this paper as [41]
In the citing paper
“Despite the challenges in collecting large-scale datasets in the pathology domain, recent pioneering works have utilized datasets ranging from 30,000 to 400,000 WSIs to train foundation models ranging in size from 28 million to 307 million parameters35–42”
From A foundation model for clinical-grade computational pathology and rare cancers detection (2024) · cites this paper as [41]
In the citing paper
“Specialty-specific foundation models20,21,23 typically focus on single imaging modalities”
From A Multimodal Vision Foundation Model for Clinical Dermatology (2025) · cites this paper as [23]
In the citing paper
“The scaling behavior observed in PanDerm’s performance aligns with recent foundation model trends20,22,23,59”
From A Multimodal Vision Foundation Model for Clinical Dermatology (2025) · cites this paper as [23]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
Scikit-learn: Machine Learning in Python
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Müller, A., Nothman, J., Louppe, G., et al. (2012) · arXiv · doi:10.48550/arxiv.1201.0490 · Cited by 71,016
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., & Guo, B. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00986 · Cited by 35,534
PyTorch: An Imperative Style, High-Performance Deep Learning Library
Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al. (2019) · arXiv · doi:10.48550/arxiv.1912.01703 · Cited by 31,473
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Advances in neural information processing systems 7
Cios, K. J., & Shields, M. E. (1997) · Neurocomputing · doi:10.1016/s0925-2312(97)90018-7 · Cited by 22,093
Least squares quantization in PCM
Lloyd, S. (1982) · IEEE Transactions on Information Theory · doi:10.1109/tit.1982.1056489 · Cited by 15,836
Goldberger, A. L., Amaral, L. A. N., Glass, L., Hausdorff, J. M., Ivanov, P. C., Mark, R. G., Mietus, J. E., Moody, G. B., Peng, C.-K., & Stanley, H. E. (2000) · Circulation · doi:10.1161/01.cir.101.23.e215 · Cited by 14,809
Research works printed but not linked: 13
- Chen, T., Kornblith, S., Norouzi, M. & Hinton, G. A simple framework for contrastive learning of visual representations. In International Conference on Machine Learning, 1597–1607 (PMLR, 2020).
- Chen, R. J. & Krishnan, R. G. Self-supervised vision transformers learn visual concepts in histopathology. In Learning Meaningful Representations of Life, NeurIPS 2021 (2022).
- Dosovitskiy, A. et al. An image is worth 16×16 words: transformers for image recognition at scale. In International Conference on Learning Representations (2021).
- Xiang, J. & Zhang, J. Exploring low-rank property in multiple instance learning for whole slide image classification. In The Eleventh International Conference on Learning Representations (2022).
- Wang, X. et al. SCL-WC: cross-slide contrastive learning for weakly-supervised whole-slide image classification. Advances in Neural Information Processing Systems 35, 18009–18021 (2022).
- Zhou, J. et al. iBOT: image BERT pre-training with online tokenizer. In International Conference on Learning Representations (2022).
- Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. BERT: pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) (2018).
- Sablayrolles, A., Douze, M., Schmid, C. & Jégou, H. Spreading vectors for similarity search. In International Conference on Learning Representations (2019).
- Touvron, H., Vedaldi, A., Douze, M. & Jegou, H. Fixing the train–test resolution discrepancy. In Advances in Neural Information Processing Systems, Vol. 32 (eds Wallach, H. et al.) (Curran Associates, 2019).
- Dao, T., Fu, D. Y., Ermon, S., Rudra, A. & Ré, C. FlashAttention: fast and memory-efficient exact attention with IO-awareness. In Advances in Neural Information Processing Systems (2022).
- Loshchilov, I. & Hutter, F. Decoupled weight decay regularization. In International Conference on Learning Representations (2019).
- Sarıyıldız, M. B., Kalantidis, Y., Alahari, K. & Larlus, D. No reason for no supervision: improved generalization in supervised models. In The Eleventh International Conference on Learning Representations (2023).
- Yang, J., Chen, H., Yan, J., Chen, X. & Yao, J. Towards better understanding and better generalization of low-shot classification in histology images with contrastive learning. In International Conference on Learning Representations (2021).
Cited by
Sorted by citations
Items 801–900 of 1,003
Zhu, Q., Lin, Y., Fu, W., Tang, W. H., Li, J., Zhang, Y., Li, B., Guo, X., Wang, F. S., Qi, H., et al. (2026) · Nature Biomedical Engineering · doi:10.1038/s41551-026-01713-8 · Cited by 2
Multimodal deep learning for papillary thyroid carcinoma diagnosis using ultrasound and cytology
Kalejahi, B. K., & Rajabi, M. J. (2026) · BMC Medical Imaging · doi:10.1186/s12880-026-02526-6 · Cited by 2
“We did not compare to recent foundation models for cytology specifically (e.g., UNI [31], ProvGigaPath [32]) because they are an order of magnitude larger and would not fit on our 8 GB GPU in fine-tuning mode.”
Mantle cell lymphoma artificial intelligence prognostic index using hematoxylin and eosin histology
Lippl, J., Reinke, S., Koch, K., Oschlies, I., Schrod, S., Wolfseher, L., Hüttl, P., Huttner, M., Beà, S., Campo, E., et al. (2026) · Leukemia · doi:10.1038/s41375-026-03036-8 · Cited by 1
“However, using at least one of these previously published foundation models (UNI [35]), we did not observe a major advantage of MAIPI compared to Prov-GigaPath [23] used in our analysis (data not shown).”
Gao, Z., Ge, J., Lu, J., Wu, L., Clayton, H., Zhang, D., Liu, J., Machado, I. P., Chen, Y., Crispin‐Ortuzar, M., et al. (2026) · Intelligent Oncology · doi:10.1016/j.intonc.2026.100074 · Cited by 1
COAST: Context-Aware Differential Learning for Gene Expression Prediction in Spatial Transcriptomics
Byeon, K., Park, S., Lim, J., & Kwak, J. T. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38224-5_6 · Cited by 1
Next-generation kidney tissue analysis — spatial omics and digital pathology
Yoshikawa, T., Bülow, R. D., Boor, P., & Yanagita, M. (2026) · Nature Reviews Nephrology · doi:10.1038/s41581-026-01102-z · Cited by 3
“Foundation models in computational pathology have been trained using image data 157 , or multimodal data sources including images and texts 158 .”
Language-Guided Segmentation of Medical Images: A Review of Foundation Models
Qamar, S. (2026) · Bioengineering · doi:10.3390/bioengineering13070803 · Cited by 1
“Vision-language and self-supervised pretraining on weakly labeled image-text pairs (PLIP, Quilt-1M, CT-RATE) and unlabeled images (DINO, DINOv2, masked image modeling) reduces this dependence [59–62, 67, 68,”
InfluentialHao, N., Yang, X., Yan, B., Li, D., Huang, J., Wu, X., Ruiz, E. S., de Luzuriaga, A. R., Zhao, C., & Wan, G. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag387 · Cited by 1
“DeepSpot2Cell [ 69] adopts a DeepSet-based architecture and leverages pretrained PFMs, whose choice could be tailored to specific tasks, such as UNI [ 70] or H-Optimus-0 [ 71].”
Hong, S., & Na, H. (2026) · Information · doi:10.3390/info17070712 · Cited by 1
Bao, W., Wang, T., Xia, X., Chen, L., Qi, H., Feng, Y., Fa, X., Ma, G., Chen, S., Leng, K., et al. (2026) · Translational Oncology · doi:10.1016/j.tranon.2026.102946 · Cited by 1
“UNI is a self-supervised visual Transformer model trained on a large histopathology corpus [12], which performs best in intra-group fitting.”
InfluentialMagisson, F., He, Z., Millward, J., Harris, A., Chen, Z., Mielke, L. A., Tran, K., Ward, R. L., Hawkins, N. J., Sieber, O. M., et al. (2026) · Gastroenterology · doi:10.1053/j.gastro.2026.07.009 · Cited by 1
MMSF: Multitask and multimodal supervised framework for WSI classification and survival analysis
She, C., Chen, C., Zhang, X., Wang, B., Liu, L., Shao, C., & Bian, Y. (2026) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2026.111045 · Cited by 1
Dong, C., Liu, Y., Yin, J., Yan, W., Sun, J., Zhang, X., Huo, S., Yu, F., & Zhou, Y. (2026) · Digital Health · doi:10.1177/20552076261474637 · Cited by 3
“68,69”
Merker, V. L., Ahlawat, S., Avery, R. A., Bradford, D., Gross, A. M., Janusz, J. A., Lessing, A. J., Manth, L., McManus, M. L., Oberlander, B., et al. (2026) · Cancers · doi:10.3390/cancers18152463 · Cited by 1
“Histology and AI-assisted digital pathology platforms can enhance diagnostic accuracy through centralized review [80,82–84].”
Su, L., Li, R., Zhang, G., Liu, Z., Geng, X., Wang, H., Gai, W., Wu, H., Tian, J., & Du, Y. (2026) · Visual Computing for Industry Biomedicine and Art · doi:10.1186/s42492-026-00227-z · Cited by 1
A protocol for evaluating robustness to H&E staining variation in computational pathology models
Schoenpflug, L. A., van den Berg, N., Andani, S., Horeweg, N., Wolf, J. B., Bosse, T., Koelzer, V. H., & Lafarge, M. W. (2026) · Journal of Pathology Informatics · doi:10.1016/j.jpi.2026.100702 · Cited by 4
“To induce realistic variability, we sampled across multiple hyperparameters: three foundation model feature extractors (UNI2-h i,38 , H-Optimus-1 j,39 , Virchow2 k,40 ) with 100 models per extractor, arbitrary random initialization seeds, two weight decay values (0 or , MSI-strati fi ed 70:30…”
DBT-DINO: Toward Foundation Model–Based Analysis of Digital Breast Tomosynthesis
Dorfner, F. J., Dorster, M. A., Connolly, R., Gentilhomme, O., Gibbs, E., Wander, S. A., Schultz, T., Bahl, M., Daye, D., Kim, A. E., et al. (2026) · Radiology · doi:10.1148/radiol.254034 · Cited by 1
Corso, F., Zec, A., Favali, M., Ligero, M., Hilgers, L., Invernizzi, L. M., Mazzeo, L., Morales, A. M., Müller‐Franzes, G. A., Machado, I. P., et al. (2026) · Cell Reports Medicine · doi:10.1016/j.xcrm.2026.102966 · Cited by 2
Kang, H., Nam, Y., Ngo, D., Cho, M., 이원애, Shin, H.-C., Park, Y., An, H. J., Kim, H., Cho, N. H., et al. (2026) · The Journal of Pathology · doi:10.1002/path.70105 · Cited by 1
“[20,25].”
Rethinking Architectural Complexity in Deep Vision Models for Histopathological Image Classification
Abbas, Q., Liaqat, M. I., Qayum, S., & Hina, S. (2026) · Journal of Imaging Informatics in Medicine · doi:10.1007/s10278-026-02175-9 · Cited by 1
Gao, G., Yan, R., Song, A. H., Hsieh, H.-C., Barner, L. A. E., Wang, F., Brenes, D., Chow, S. S. L., Wang, R., Bishop, K. W., et al. (2026) · Nature Biomedical Engineering · doi:10.1038/s41551-026-01760-1 · Cited by 7
Self-citationPractical applications of computational pathology: bridging technology and clinical practice
Hanna, M. G., Rashidi, H., & Pantanowitz, L. (2026) · Pathology · doi:10.1016/j.pathol.2026.06.012 · Cited by 1
Artificial Intelligence in equivocal prostatic needle biopsies: benign mimics and cancer detection
Bawahab, A. A. (2026) · Archivio Italiano di Urologia e Andrologia · doi:10.4081/aiua.2026.15673 · Cited by 1
Aswolinskiy, W., Motiwala, M., Paulikat, M., Nelius, N., Ahadova, A., Adam, P., Schmidt, D., Kloor, M., & Aichmüller, C. (2026) · npj Precision Oncology · doi:10.1038/s41698-026-01581-0 · Cited by 1
Ensemble learning of pathology foundation models for precision oncology
Luo, X., Wang, X., Eweje, F. R., Zhang, X., Marti, J. L. G., Cascarino, S., Yang, S., Li, Y., Quinton, R. J., Xiang, J., et al. (2026) · Cancer Cell · doi:10.1016/j.ccell.2026.08.008 · Cited by 17
Neamțu, A.-A., Barna, R. A., Vigdorovits, A., Hotinceanu, I. A., Muresan, M. M., Paşcalău, A. V., & Pop, O.-L. (2026) · Cancers · doi:10.3390/cancers18182974 · Cited by 1
Melanoma Prognostication Using AI ‐Guided Histopathology
Lee, E., Karagenova, R., & Semenov, Y. R. (2026) · International Journal of Dermatology · doi:10.1111/ijd.70585 · Cited by 2
“Early architectures relied on general-purpose CNNs pretrained on natural image datasets such as ImageNet (e.g., ResNet, VGG) [8], which demonstrated proof- of-concept utility, but were not optimized for the unique visual properties of histopathological tissue [9].”
Discovering Latent Scientific Concepts Through Discrete Representation Learning
Stoimchev, M., Verma, M., Filliol, A., Skamagki, M., Flowers, S., Cohen, J., Azadian, Z., Newlin, N., Mottaleb, M. S. A., Džeroski, S., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-41086-3_12 · Cited by 1
Transfer learning may explain pigeons’ ability to detect cancer in histopathology
Kilim, O., Báskay, J., Biricz, A., Bedőházi, Z., Pollner, P., & Csabai, I. (2024) · Bioinspiration & Biomimetics · doi:10.1088/1748-3190/ad6825 · Cited by 2
“To add a further baseline to compare these results to, we repeated these experiments with a histopathology pre-trained, state-of-the-art model UNI [16].”
Application of artificial intelligence in immuno-oncology
Monfort‐Lanzas, P., Gronauer, R., Balaz, M., Floriani, G., & Hackl, H. (2024) · Elsevier eBooks · doi:10.1016/b978-0-443-14064-8.00017-5 · Cited by 1
Tang, L.-Z., Zhang, Z., Yang, J., Feng, Y., Wang, Q., Sun, S., & Shao, H. (2025) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2025.3592359 · Cited by 2
“3) Implementation Details: The slide patching used CLAM [17], and the feature extractor chosen for patch feature extraction was Uni [50].”
DeepPathway: predicting pathway expression from histopathology images
Ahsan, M., Hanley, K. P., Fergie, M., O’Leary, C., Borst, G. R., Roncaroli, F., Minhas, F., Rattray, M., Iqbal, M., & Baker, S. M. (2026) · Bioinformatics · doi:10.1093/bioinformatics/btag643 · Cited by 1
“We have used the first version of optimus foundation model, i.e. H-optimus-0 , however, there are other foundation models (Chen et al. 2024, Vorontsov et al. 2024) along with H-optimus-1 (https://huggingface.co/bioptimus/H-optimus-0-1) (trained on histology images sampled from over 1 million slides…”
Organoid-ICLIP: Class Imbalance-Aware Vision-Language Learning for Organoid Mitosis Classification
Sohail, A., Deb, O., & Ulhaq, A. (2025) · Proceedings - International Conference on Image Processing · doi:10.1109/icip55913.2025.11084740 · Cited by 1
Redekop, E., Pleasure, M., Ivezić, V., Wang, Z., Flores, K., Sisk, A. E. J., Speier, W. F., & Arnold, C. W. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00514 · Cited by 1
Dawood, M., Thomas, E. S., Cooper, R. A., Pescia, C., Sozanska, A. M., Ryou, H., Royston, D. J., & Rittscher, J. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-05141-7_58 · Cited by 1
Enhancing zero-shot brain tumor subtype classification via fine-grained patch-text alignment
Gan, L., Zhang, J., Qu, L., Wang, Y., Wu, S., & Sun, X. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.130161 · Cited by 3
Notton, C., Sharma, V., Trinh, V. Q., Chen, L., Xu, M., Varma, S., & Hosseini, M. S. (2026) · Medical Image Analysis · doi:10.1016/j.media.2026.104004 · Cited by 1
Prompt-guided foundation model tuning for pathology image classification
Lin, Y., Zhu, Z., Cheng, K.-T. T., & Chen, H. (2026) · Medical Image Analysis · doi:10.1016/j.media.2026.104214 · Cited by 4
A Framework for Evaluating the Efficacy of Foundation Embedding Models in HealthcarePreprint
Xu, S. H., Gui, H., Rotemberg, V. M., Wang, T., Chen, Y. T., & Daneshjou, R. (2024) · medRxiv · doi:10.1101/2024.04.17.24305983 · Cited by 1
“Finally, we aim to extend and assess this framework in other medical fields such as cardiology, radiology, and pathology, where there is an abundance of multimodal data and foundation models are beginning to show promise (Huang et al., 2023; Zhang et al., 2023; Chen et al., 2024).”
BackgroundHIBRID: Histology and ct-DNA based Risk-stratification with Deep LearningPreprint
Loeffler, C. M. L., Bando, H., Sainath, S., Muti, H. S., Jiang, X., van Treeck, M., Reitsam, N. G., Carrero, Z. I., Nishikawa, T., Misumi, T., et al. (2024) · medRxiv · doi:10.1101/2024.07.23.24310822 · Cited by 1
Lu, C. Y., Xu, H., Wang, M., Shi, D., Qin, H., Madabhushi, A., Gao, P., & Cong, F. (2024) · Research Square · doi:10.21203/rs.3.rs-4704418/v1 · Cited by 1
Carrillo‐Pérez, F., Cramer, E. M., Pizurica, M., Andor, N., & Gevaert, O. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.08.19.608555 · Cited by 1
“Thus, similarly to what other researchers have reported in relevant cancer tasks, using a SSL trained model on digital pathology as a feature extractor outperforms feature extractors trained on natural images for ploidy quantification 20,21 .”
BackgroundResultChen, C., Yeh, Y., Lin, M. M., & Yeh, C. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.10.03.616445 · Cited by 1
Contrastive Learning for Omics-guided Whole-slide Visual Embedding RepresentationPreprint
Yu, S., Kim, Y., Kim, H., Lee, S., & Kim, K. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.01.12.632280 · Cited by 2
An Efficient and Interpretable Foundation Model for Retinal Image Analysis in Disease DiagnosisPreprint
Dai, W., Chen, Z. J., Yao, Y., Chen, Y., Fang, J., Bai, Q., Xu, C., Wu, H., Ding, H., Yang, H., et al. (2025) · medRxiv · doi:10.1101/2025.02.19.25322447 · Cited by 1
Bitto, V., Jiang, X., Baumann, M. A., Kather, J. N., & Kurth, I. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.03.31.646351 · Cited by 1
Schuiveling, M., van Duin, I. A. J., ter Maat, L. S., van der Weerd, J. C., Verheijden, R. J., van den Berkmortel, F. W. P. J., Blank, C. U., Breimer, G. E., Burgers, F. H., Boers‐Sonderen, M. J., et al. (2025) · medRxiv · doi:10.1101/2025.05.28.25328410 · Cited by 1
Thiesen, A. H., Domanskyi, S., pour, A. F., Zhang, J., Sheridan, T., Neuhauser, S. B., Stetson, A. E., Dannheim, K., Cameron, D. B., Ahn, S. S., et al. (2025) · medRxiv · doi:10.1101/2025.06.10.25328700 · Cited by 1
“0.958, respectively), outperforming CTransPath (p 5x10 -4 ) and InceptionV3 (p 5x10 -5 ), with no significant difference between UNI and CONCH themselves Supplementary Figure S4 .”
InfluentialPatkar, S., Rosean, T. R., Patel, P. G., Harmon, S. A., Choyke, P. L., Jamaspishvili, T., & Türkbey, B. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.06.11.658673 · Cited by 2
ODFormer: a Virtual Organoid for Predicting Personalized Therapeutic Responses in Pancreatic CancerPreprint
Xu, J., Yang, X., Li, Y., Wang, H., Li, Y., Tang, S., Guo, S., Zou, J., Shi, X., Chen, Z., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.07.08.663664 · Cited by 2
Akan, T., Gelir, F., Aishwarya, R., Bhuiyan, M. S., & Bhuiyan, M. A. N. (2025) · Research Square · doi:10.21203/rs.3.rs-6710077/v1 · Cited by 1
“In the context of musculoskeletal disease diagnosis, this approach is particularly pertinent, as it is imperative to accurately evaluate subtle histopathological changes in order to provide timely and effective intervention [15, 16].”
Zhuang, P., Cheng, S., Zhao, Y., Kang, H., Sun, C., Li, J., Zheng, W., Zhang, L., Wang, X., Dai, J., et al. (2025) · Research Square · doi:10.21203/rs.3.rs-7397345/v1 · Cited by 1
Maleki, D., Shaikh, N., Li, X., Nie, Y., Venugopal, R., & Kurkure, U. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.12.675923 · Cited by 1
“For example, UNI2’s strength in diverse, fine-grained ROI classification could be attributed to its pretraining on a large, multi-institutional dataset that fosters robust feature learning 16 .”
Designing smart spatial omics experiments with S2OmicsPreprint
Yuan, M., Jin, K., Yan, H., Schroeder, A. R., Luo, C., Yao, S., Domoulin, B., Levinsohn, J. L., Luo, T., Clemenceau, J. R. V., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.21.677634 · Cited by 1
Nonchev, K., Manaiev, G., Koelzer, V. H., & Rätsch, G. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.23.678121 · Cited by 3
“Notable examples include UNI [17], Phikon-v2 [18], and H-Optimus-0 [19], which mostly rely on vision transformers (ViT) to learn high-dimensional morphological representations.”
Evaluating Foundation Models with Pathological Concept Learning for Kidney CancerPreprint
Gao, S., Wang, S., Gao, Y., Wang, B., Zhuang, X., Warren, A. Y., Stewart, G. D., Jones, J. O., & Crispin‐Ortuzar, M. (2025) · medRxiv · doi:10.1101/2025.09.29.25336908 · Cited by 1
“Foundation models have significantly enhanced the ability of computational pathology frameworks to make clinically relevant predictions [5, 7, 6].”
Tajeddin, N. Z., Jahanifar, M., Eastwood, M., Gunesli, G. N., Arends, M. J., & Rajpoot, N. (2025) · medRxiv · doi:10.1101/2025.09.28.25336824 · Cited by 1
“…field has rapidly moved toward large pathology “foundation models” pre-trained on millions of tiles or whole-slide images [21]; models such as UNI [22], Virchow [23], and Prov-GigaPath [24] have achieved state-of-the-art results on a broad suite of downstream tasks and have demonstrated strong…”
InfluentialSystematic mapping of human tissue microanatomy reveals age-associated remodeling and resiliencePreprint
Buljan, I., Bagó-Horváth, Z., & Rendeiro, A. F. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.10.13.682110 · Cited by 1
Farndale, L., Rakovic, K., Powley, I. R., Hatthakarnkul, P., Seyedshahi, F. A., Jesenakova, S., Santana, L. P. S., Martinelli, S., Ballantyne, F., Officer-Jones, L., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.10.16.682563 · Cited by 1
“SSL applications in pathology have focused on brightfield images [8, 9, 10, 11], but the potential to discover high-order, cross-protein spatial biomarkers in full mIF datasets remains un-tapped.”
Quantifying predictability of gene expression from histology imagePreprint
Xia, C., Yao, J., & Gao, G. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.11.04.686651 · Cited by 1
REECAP: Contrastive learning of retinal aging reveals genetic loci linking morphology to eye diseasePreprint
Shilova, L., Sens, D., Aliyeva, A., Chaudhary, S., Xu, Q., Salin, E., Schiefelbein, J. B., Asani, B., Amarie, O. V., Schneltzer, E., et al. (2025) · medRxiv · doi:10.1101/2025.11.19.25340555 · Cited by 2
Modeling patient tissues at molecular resolution with EvaPreprint
Liu, Y., Sharma, R., Bieniosek, M. F., Kang, A., Wu, E., Chou, P., Li, I., Rahim, M. K., Bauer, E., Ji, R., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.10.693553 · Cited by 3
“Prior work has extensively explored applications of self-supervised vision transformer (ViT) and masked autoencoder (MAE) [10–12] for digital pathology [7,13,14].”
InfluentialDevelopment of the Early Childhood Duodenum across Ancestry, Geography and EnvironmentPreprint
Casal, J. D. S., Raghunathan, K., Asare, C., Plone, A., Moradinasab, N., Iqbal, J., Wood, L., Ngwa, E. M., Chen, X., Rhoads, S. F., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.15.694427 · Cited by 2
Single-cell spatial multi-omics molecular pathology enabled by SuperFocusPreprint
Lu, Y., Tian, X., Vicari, M., Enninful, A., Bao, S., Bai, Z., Liu, C., Zhang, X., Andrén, P. E., Lundeberg, J., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.26.696575 · Cited by 1
“Recent whole-slide pathology foundation models such as UNI (63) and Prov-GigaPath (64), together with ROI-selection frameworks such as S2-omics (65), make this combined strategy increasingly realistic.”
Generative AI Enables Breast Cancer Genomic Subtype Prediction from Histology ImagesPreprint
Simon, B. G., Weiß, C. L., Chan, D., Mangiante, L., Ma, Z., Smith, N. H., Meisner, A., Rae, J. M., Speers, C. W., Albain, K. S., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.29.692457 · Cited by 1
“UNI2 was used to extract til e embeddings according to the CLAM pipeline 8,22 .”
miniMTI: minimal multiplex tissue imaging enhances biomarker expression prediction from histologyPreprint
Chang, Y. H., Sims, Z., Govindarajan, S., Ait-Ahmad, K., Ak, Ç., Kuykendall, M., Mills, G. B., & Eksi, S. E. (2026) · Research Square · doi:10.21203/rs.3.rs-8865586/v1 · Cited by 1
“As a representative late-fusion baseline, we combined UNI 39 , a general-purpose histopathology foundation model trained on large-scale H&E data, with KRONOS 40 , a spatial proteomics foundation model pre-trained on MTI.”
Schroeder, A. R., Yu, X., Li, W., Yuan, M., Mao, L., Yang, J., Sachs, N., Dumoulin, B., Xu, G., Luo, X., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.01.30.702675 · Cited by 1
Ayad, M. A., McCortney, K., Congivaram, H. T. S., Hjerthen, M. G., Steffens, A., Zhang, H., Youngblood, M. W., Heimberger, A. B., Chandler, J. P., Jamshidi, P., et al. (2026) · medRxiv · doi:10.64898/2026.02.23.26346491 · Cited by 1
A spatial multi-omic portrait of survival outcome for clear cell renal cell carcinomaPreprint
Meyer, L., Engler, S., Lutz, M., Schraml, P., Rutishauser, D., Bertolini, A., Lienhard, M., Beisel, C., Singer, F., de Souza, N., et al. (2026) · medRxiv · doi:10.64898/2026.03.02.26347390 · Cited by 1
Toward Computationally Complete Spatial OmicsPreprint
Li, W., Mao, L., Liu, Y., Peng, F., Sachs, N., Wu, W., Yiu, S. P. T., Yan, H., Schroeder, A. R., Yu, X., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.03.03.709303 · Cited by 1
GenBio-PathFM: A State-of-the-Art Foundation Model for HistopathologyPreprint
Kapse, S., Aygün, M., Cole, E., Lundberg, E. K., Song, L., & Xing, E. P. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.03.17.712534 · Cited by 1
“Modern histopathology requires the integration of complex, multi-scale morphological signals to perform tasks ranging from tumor subtyping and grading to molecular state inference [1–4].”
H2O: A Foundation Model Bridging Histopathology to Spatial Multi-Omics ProfilingPreprint
Gu, Y., Wu, Z., Yan, R., Wang, Z., Li, Y., Lin, S., Cui, Y., Lai, H., Luo, X., Zhou, S. K., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.04.21.717342 · Cited by 2
A multimodal foundation model to integrate histology with spatial molecular programsPreprint
Zhang, Z., Qin, B., Zhao, Y., Qi, Z., Xu, H., Wang, Y., Zheng, W., Dai, J., Chen, A., Wang, H., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.06.01.729028 · Cited by 1
Ali, P. J. M., Vincent, N., Abdulla, S. S., Fadhl, H. N. M., Blilie, A., Szolnoky, K., Mielcarz, J. A., Ji, X., Kartasalo, K., Al‐Talabani, A. K., et al. (2026) · Scientific Data · doi:10.1038/s41597-026-07798-9 · Cited by 2
“Digital pathology and artificial”
Are Stronger Feature Representations All You Need for Histology Image Search?
Chen, R. J., & Mahmood, F. (2024) · NEJM AI · doi:10.1056/aie2400314
Self-citationBrattoli, B., Mostafavi, M., Lee, T., Jung, W., Ryu, J., Park, S., Park, J., Pereira, S., Shin, S.-H., Choi, S., et al. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4857105
VIB Hackathon on spatial omics tools and methodsPreprint
Rombaut, B., Pollaris, L., Sang-aram, C., Cruysse, M. V., Cannoodt, R., Vernaillen, F., Defauw, A., Mortier, J., Ali, M., Bestak, K., et al. (2024) · doi:10.37044/osf.io/uf7d3
Sahm, F. (2024) · Neuro-Oncology · doi:10.1093/neuonc/noae124
Wang, R., Gunesli, G. N., Skingen, V. E., Valen, K.-A. F., Lyng, H., Young, L. S., & Rajpoot, N. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.08.16.608264
“UNI is a domain-specific foundation model for H&E-stained histology images.”
MethodResultInfluentialBenchmarking Vision Encoders for Survival Analysis using Histopathological ImagesPreprint
Nizami, A., & Halder, A. (2024) · medRxiv · doi:10.1101/2024.08.23.24312362
“UNI: UNI [22] is a general-purpose self-supervised model trained using more than 100 million images from 100k WSIs across 20 major organ types, making it the largest pretrained vision encoder for histopathology.”
MethodChen, J. H., Elmelech, L., Tang, A. L., & Hacohen, N. (2024) · Current Opinion in Immunology · doi:10.1016/j.coi.2024.102463
Mehrabian, H., Brodbeck, J., Lyu, P., Vaquero, E., Aggarwal, A., & Diehl, L. J. (2024) · Scientific Reports · doi:10.1038/s41598-024-69244-3 · Cited by 2
Machine-learning convergent melanocytic morphology despite noisy archival slidesPreprint
Tada, M., Gaskins, G., Ghandian, S., Mew, N. C., Keiser, M. J., & Keiser, E. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.09.12.612732
“Computationally, compared to some large-scale studies on tens to hundreds of thousands of private multi-tissue slides 15,51 , this melanocytic tissue dataset is small, varied, and visually noisy.”
ResultNavigating the Future of Healthcare
Musunuri, R. L., & Bhatt, A. (2024) · Advances in healthcare information systems and administration book series · doi:10.4018/979-8-3693-8552-4.ch003
Hieber, D., Holl, F., Nickl, V., Liesche‐Starnecker, F., & Schobel, J. (2024) · IntechOpen eBooks · doi:10.5772/intechopen.1007366
Multi-stain modelling of histopathology slides for breast cancer prognosis predictionPreprint
Sharma, A., Gustafsson, F., Hartman, J., & Rantalainen, M. (2024) · medRxiv · doi:10.1101/2024.11.10.24317066 · Cited by 1
“UNI is a foundation model trained using a self-supervised learning method called DINOv2 [30] on 100k H&E WSIs across multiple tissue types [28].”
Carretero, I., Meseguer, P., del Amor, R., & Naranjo, V. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-77731-8_36
MECFormer: Multi-task Whole Slide Image Classification with Expert Consultation Network
Bui, D. C., & Kwak, J. T. (2024) · Lecture notes in computer science · doi:10.1007/978-981-96-0901-7_5 · Cited by 1
Zheng, Q., Jiao, P., Yang, R., Fan, J., Liu, Y., Yang, X., Yuan, J., Chen, Z., & Liu, X. (2025) · World Journal of Urology · doi:10.1007/s00345-025-05440-8
Nouyed, M. I., Doretto, G., & Adjeroh, D. (2024) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm62325.2024.10822778
SS-WSSS: Small-Scale Weakly Supervised Semantic Segmentation for Histopathology Image
Lin, J., Han, G., Zhou, J., Shi, Z., Liu, Z., & Han, C. (2024) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm62325.2024.10821867
Peng, 费. F., Wentian, 思. S., & Minchao, 张. Z. (2024) · Acta Optica Sinica (Online) · doi:10.3788/aosol240448
Dacic, S., Abel, J., Hennek, S. R., Travis, W. D., & Wistuba, I. I. (2025) · Journal of Thoracic Oncology · doi:10.1016/j.jtho.2024.11.024
Jimenez-Martin, L., Hernández-Pérez, C., & Vilaplana, V. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-77789-9_16
Replicability of large-scale studies in computational pathology
Chamaria, D., Han, W., McClintock, D. S., & Tavolara, T. E. (2025) · SPIE Medical Imaging: Digital and Computational Pathology · doi:10.1117/12.3047510
“In no particular order, these include PRISM [5], Virchow [6], CONCH [7], UNI [8], and Prov-GigaPath [9].”
Ganz, J., Ammeling, J., Rosbach, E., Lausser, L. M., Bertram, C., Breininger, K., & Aubreville, M. (2025) · Informatik aktuell · doi:10.1007/978-3-658-47422-5_15
Efficient Multi-Stage Self-Supervised Learning for Pathology Image Analysis via Masking
Feng, M. Y., Huang, W., & Hu, L. Z. (2025) · IEEE Access · doi:10.1109/access.2025.3556435
“Recently, DINOv2 [36] has become the mainstream pre-training model in the field of pathology [12], [13].”
Optimization of transfer learning of foundation models for hyperspectral histologic imaging
Hellman, M. D., Ma, L., Yu, J. J. Q., & Fei, B. (2025) · SPIE Medical Imaging: Digital and Computational Pathology · doi:10.1117/12.3048965
“These models also exhibit promising few-shot performance metrics which make them strong candidates for application in domains where data is not widely available 5 .”
Winzell, F., Arvidsson, I., Åström, K., Overgaard, N. C., Marginean, F.-E., Simoulis, A., Bjartell, A. S., Krzyzanowska, A., & Heyden, A. (2025) · SPIE Medical Imaging: Digital and Computational Pathology · doi:10.1117/12.3046323
“36”
InfluentialChen, Q., Wang, Z., Xia, L., Shi, Y., Xu, B., Chai, J., Zhang, T., & Wang, C. (2025) · medRxiv · doi:10.1101/2025.04.13.25323912
Metadata
- Authors
- Chen, R. J., Ding, T., Lu, M. Y., Williamson, D. F. K., Jaume, G., Song, A. H., Chen, B., Zhang, A., Shao, D., Shaban, M., Williams, M., Oldenburg, L., Weishaupt, L. L., Wang, J. J., Vaidya, A., Le, L. P., Gerber, G., Sahai, S., Williams, W., & Mahmood, F.
- Affiliations
- Broad Institute (US)
- Brigham and Women's Hospital (US)
- Harvard University (US)
- Corresponding author
- Mahmood, Faisal
- ORCID
- Venue
- Nature Medicine
- Published
- 2024-03
- Volume
- 30
- Issue
- 3
- Pages
- 850-862
- ISSN
- 1078-8956 (ISSN-L)
- 1546-170X
- PMC
- PMC11403354
- OpenAlex
- W4392947521
- Semantic Scholar
- cac496ddebbff6bbba78242780cde01bd962bfc9
- Cited by
- 1,560
- References
- 170
- Citations by year
- 2026: 789 · 2025: 613 · 2024: 154
- FWCI
- 229.15
- Type
- journal article
- Language
- en
- Open access
- green
- www.ncbi.nlm.nih.gov
- Licence
- Locations
- Nature Medicine (publishedVersion)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- Funders
- National Human Genome Research Institute (T32HG002295)
- National Cancer Institute (T32CA251062)
- National Institute of General Medical Sciences (R35GM149270)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- computational pathology, whole slide images, self-supervised learning, foundation models, unsupervised feature learning, tissue classification, few-shot learning, cancer subtyping, histopathology, transfer learning
- MeSH terms
- Artificial Intelligence, Workflow
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
