Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning


A principal challenge in the analysis of tissue imaging data is cell segmentation—the task of identifying the precise boundary of every cell in an image. To address this problem we constructed TissueNet, a dataset for training segmentation models that contains more than 1 million manually labeled cells, an order of magnitude more than all previously published segmentation training datasets. We used TissueNet to train Mesmer, a deep-learning-enabled segmentation algorithm. We demonstrated that Mesmer is more accurate than previous methods, generalizes to the full diversity of tissue types and imaging platforms in TissueNet, and achieves human-level performance. Mesmer enabled the automated extraction of key cellular features, such as subcellular localization of protein signal, which was challenging with previous approaches. We then adapted Mesmer to harness cell lineage information in highly multiplexed datasets and used this enhanced version to quantify cell morphology changes during human gestation. All code, data and models are released as a community resource.

This is a preview of subscription content

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Fig. 1: A human-in-the-loop approach enables scalable, pixel-level annotation of large image collections.
Fig. 2: Mesmer delivers accurate nuclear and whole-cell segmentation in multiplexed images of tissues.
Fig. 3: Mesmer performs whole-cell segmentation across tissue types and imaging platforms with human-level accuracy.
Fig. 4: Mesmer enables accurate analysis of multiplex imaging data.
Fig. 5: Lineage-aware segmentation enables morphological profiling of cells in the decidua during human pregnancy.
Fig. 6: Cloud-native and on-premise software facilitates deployment of Mesmer.

Data availability

The TissueNet dataset is available at for noncommercial use.

Code availability

All software for dataset construction, model training, deployment and analysis is available on our github page All code to generate the figures in this paper is available at


  1. 1.

    Giesen, C. et al. Highly multiplexed imaging of tumor tissues with subcellular resolution by mass cytometry. Nat. Methods 11, 417–422 (2014).

    CAS  PubMed  Google Scholar 

  2. 2.

    Keren, L. et al. MIBI-TOF: a multiplexed imaging platform relates cellular phenotypes and tissue structure. Sci. Adv. 5, eaax5851 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  3. 3.

    Huang, W., Hennrick, K. & Drew, S. A colorful future of quantitative pathology: validation of Vectra technology using chromogenic multiplexed immunohistochemistry and prostate tissue microarrays. Hum. Pathol. 44, 29–38 (2013).

    CAS  PubMed  Google Scholar 

  4. 4.

    Lin, J.-R. et al. Highly multiplexed immunofluorescence imaging of human tissues and tumors using t-CyCIF and conventional optical microscopes. eLife 7, e31657 (2018).

    PubMed  PubMed Central  Google Scholar 

  5. 5.

    Gerdes, M. J. et al. Highly multiplexed single-cell analysis of formalin-fixed, paraffin-embedded cancer tissue. Proc. Natl Acad. Sci. 110, 11982–11987 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  6. 6.

    Goltsev, Y. et al. Deep profiling of mouse splenic architecture with CODEX multiplexed imaging. Cell 174, 968–981.e15 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  7. 7.

    Chen, K. H., Boettiger, A. N., Moffitt, J. R., Wang, S. & Zhuang, X. Spatially resolved, highly multiplexed RNA profiling in single cells. Science 348, aaa6090 (2015).

    PubMed  PubMed Central  Google Scholar 

  8. 8.

    Lee, J. H. et al. Highly multiplexed subcellular RNA sequencing in situ. Science 343, 1360–1363 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  9. 9.

    Moffitt, J. R. et al. Molecular, spatial and functional single-cell profiling of the hypothalamic preoptic region. Science 362, eaau5324 (2018).

    PubMed  PubMed Central  Google Scholar 

  10. 10.

    Wang, X. et al. Three-dimensional intact-tissue sequencing of single-cell transcriptional states. Science 361, eaat5691 (2018).

    PubMed  PubMed Central  Google Scholar 

  11. 11.

    Lubeck, E., Coskun, A. F., Zhiyentayev, T., Ahmad, M. & Cai, L. Single-cell in situ RNA profiling by sequential hybridization. Nat Methods 11, 360–361 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  12. 12.

    Eng, C.-H. L. et al. Transcriptome-scale super-resolved imaging in tissues by RNA seqFISH+. Nature 568, 235–239 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  13. 13.

    Rozenblatt-Rosen, O. et al. The human tumor atlas network: charting tumor transitions across space and time at single-cell resolution. Cell 181, 236–249 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  14. 14.

    Snyder, M. P. et al. The human body at cellular resolution: the NIH Human Biomolecular Atlas Program. Nature 574, 187–192 (2019).

    Google Scholar 

  15. 15.

    Regev, A. et al. The human cell atlas white paper. Preprint at (2018).

  16. 16.

    Keren, L. et al. A structured tumor-immune microenvironment in triple negative breast cancer revealed by multiplexed ion beam imaging. Cell 174, 1373–1387.e19 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Milo, R. & Phillips, R. Cell Biology by the Numbers 1st edn (Garland Science, 2015).

  18. 18.

    Mescher, A. Junqueira’s Basic Histology: Text and Atlas 13th edn (McGraw Hill, 2013).

  19. 19.

    McQuin, C. et al. CellProfiler 3.0: next-generation image processing for biology. PLoS Biol. 16, e2005970 (2018).

    PubMed  PubMed Central  Google Scholar 

  20. 20.

    Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).

    CAS  Google Scholar 

  21. 21.

    Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  22. 22.

    Berg, S. et al. ilastik: interactive machine learning for (bio)image analysis. Nat. Methods 16, 1226–1232 (2019).

    CAS  PubMed  Google Scholar 

  23. 23.

    de Chaumont, F. Icy: an open bioimage informatics platform for extended reproducible research. Nat. Methods 9, 690–696 (2012).

    CAS  PubMed  Google Scholar 

  24. 24.

    Belevich, I., Joensuu, M., Kumar, D., Vihinen, H. & Jokitalo, E. Microscopy image browser: a platform for segmentation and analysis of multidimensional datasets. PLoS Biol. 14, e1002340 (2016).

    PubMed  PubMed Central  Google Scholar 

  25. 25.

    Ronneberger, O., Fischer, P. & Brox, T. in Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 (eds Navab, N. et al.) 234–241 (Lecture Notes in Computer Science 9351, Springer, 2015).

  26. 26.

    Valen, D. A. V. et al. Deep learning automates the quantitative analysis of individual cells in live-cell imaging experiments. PLoS Comput. Biol. 12, e1005177 (2016).

    PubMed  PubMed Central  Google Scholar 

  27. 27.

    Caicedo, J. C. et al. Nucleus segmentation across imaging experiments: the 2018 Data Science Bowl. Nat. Methods 16, 1247–1253 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  28. 28.

    Stringer, C., Wang, T., Michaelos, M. & Pachitariu, M. Cellpose: a generalist algorithm for cellular segmentation. Nat. Methods 18, 100–106 (2021).

    CAS  PubMed  Google Scholar 

  29. 29.

    Hollandi, R. et al. nucleAIzer: a parameter-free deep learning framework for nucleus segmentation using image style transfer. Cell Syst. 10, 453–458.e6 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  30. 30.

    Koyuncu, C. F., Gunesli, G. N., Cetin-Atalay, R. & Gunduz-Demir, C. DeepDistance: a multi-task deep regression model for cell detection in inverted microscopy images. Med. Image Anal. 63, 101720 (2020).

    PubMed  Google Scholar 

  31. 31.

    Yang, L. et al. NuSeT: A deep learning tool for reliably separating and analyzing crowded cells. PLoS Comput. Biol. 16, e1008193 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  32. 32.

    Yu, W. et al. CCDB:6843, mus musculus, Neuroblastoma. CIL. Dataset.

  33. 33.

    Koyuncu, C. F., Cetin‐Atalay, R. & Gunduz‐Demir, C. Object‐oriented segmentation of cell nuclei in fluorescence microscopy images. Cytometry A 93, 1019–1028 (2018).

    PubMed  Google Scholar 

  34. 34.

    Ljosa, V., Sokolnicki, K. L. & Carpenter, A. E. Annotated high-throughput microscopy image sets for validation. Nat. Methods 9, 637–637 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  35. 35.

    Kumar, N. et al. A multi-organ nucleus segmentation challenge. IEEE Trans. Med. Imaging 39, 1380–1391 (2020).

    PubMed  Google Scholar 

  36. 36.

    Verma, R. et al. MoNuSAC2020: A Multi-organ Nuclei Segmentation and Classification Challenge. IEEE Trans. Med. Imaging 10.1109/TMI.2021.3085712 (2021).

  37. 37.

    Moen, E. et al. Accurate cell tracking and lineage construction in live-cell imaging experiments with deep learning. Preprint at bioRxiv (2019).

  38. 38.

    Gamper, J. et al. PanNuke dataset extension, insights and baselines. Preprint at (2020).

  39. 39.

    Bannon, D. et al. DeepCell Kiosk: scaling deep learning–enabled cellular image analysis with Kubernetes. Nat. Methods 18, 43–45 (2021).

    CAS  PubMed  Google Scholar 

  40. 40.

    Haberl, M. G. et al. CDeep3M—plug-and-play cloud-based deep learning for image segmentation. Nat. Methods 15, 677–680 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41.

    Ouyang, W., Mueller, F., Hjelmare, M., Lundberg, E. & Zimmer, C. ImJoy: an open-source computational platform for the deep learning era. Nat. Methods 16, 1199–1200 (2019).

    CAS  PubMed  Google Scholar 

  42. 42.

    von Chamier, L. et al. Democratising deep learning for microscopy with ZeroCostDL4Mic. Nat. Commun. 12, 2276 (2021).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Hughes, A. J. et al. a tool for rapid, flexible, crowd-based annotation of images. Nat. Methods 15, 587–590 (2018).

    CAS  PubMed  Google Scholar 

  44. 44.

    Ouyang, W., Le, T., Xu, H. & Lundberg, E. Interactive biomedical segmentation tool powered by deep learning and ImJoy. F1000Research 10, 142 (2021).

    Google Scholar 

  45. 45.

    Wolny, A. et al. Accurate and versatile 3D segmentation of plant tissues at cellular resolution. eLife 9, e57613 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  46. 46.

    DeepCell Label:

  47. 47.

    Lin, T.-Y. et al. Feature pyramid networks for object detection. In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2117–2125 (IEEE, 2017).

  48. 48.

    Tan, M., Pang, R. & Le, Q. V. EfficientDet: scalable and efficient object detection. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 10778–10787 (IEEE, 2020).

  49. 49.

    He, K., Zhang, X., Ren, S. & Sun, J. Deep residual learning for image recognition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 770–778 (IEEE, 2016).

  50. 50.

    Zuiderveld, K. in Graphics Gems (ed Heckbert, P. S.) Ch. VIII.5 (Academic Press, 1994).

  51. 51.

    Chevalier, G. Make smooth predictions by blending image patches, such as for image segmentation.

  52. 52.

    Meyer, F. & Beucher, S. Morphological segmentation. J. Vis. Commun. Image R 1, 21–46 (1990).

    Google Scholar 

  53. 53.

    Weigert, M., Schmidt, U., Haase, R., Sugawara, K. & Myers, G. Star-convex polyhedra for 3D object detection and segmentation in microscopy. In IEEE Winter Conference on Applications of Computer Vision (WACV) 3655–3662 (IEEE, 2020).

  54. 54.

    Fu, C.-Y., Shvets, M. & Berg, A. C. RetinaMask: learning to predict masks improves state-of-the-art single-shot detection for free. Preprint at (2019).

  55. 55.

    Schürch, C. M. et al. Coordinated cellular neighborhoods orchestrate antitumoral immunity at the colorectal cancer invasive front. Cell 182, 1341–1359.e19 (2020).

    PubMed  PubMed Central  Google Scholar 

  56. 56.

    Ali, H. R. et al. Imaging mass cytometry and multiplatform genomics define the phenogenomic landscape of breast cancer. Nat. Cancer 1, 163–175 (2020).

    Google Scholar 

  57. 57.

    Gaglia, G. et al. HSF1 phase transition mediates stress adaptation and cell fate decisions. Nat. Cell Biol. 22, 151–158 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  58. 58.

    Nelson, D. E. et al. Oscillations in NF-κB signaling control the dynamics of gene expression. Science 306, 704–708 (2004).

    CAS  PubMed  Google Scholar 

  59. 59.

    Kumar, K. P., McBride, K. M., Weaver, B. K., Dingwall, C. & Reich, N. C. Regulated nuclear-cytoplasmic localization of interferon regulatory factor 3, a subunit of double-stranded RNA-activated factor 1. Mol. Cell Biol. 20, 4159–4168 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  60. 60.

    Wolff, A. C. et al. Recommendations for human epidermal growth factor receptor 2 testing in Breast Cancer: American Society of Clinical Oncology/College of American pathologists clinical practice guideline update. J. Clin. Oncol. 31, 3997–4013 (2013).

    PubMed  Google Scholar 

  61. 61.

    Risom, T. et al. Transition to invasive breast cancer is associated with progressive changes in the structure and composition of tumor stroma. Preprint at bioRxiv (2021)

  62. 62.

    Ark Analysis.

  63. 63.

    Koss, L. G. Diagnostic Cytology and Its Histopathologic Bases. (J.B. Lippincott Company, 1979).

  64. 64.

    Erlebacher, A. Immunology of the maternal-fetal interface. Annu. Rev. Immunol. 31, 387–411 (2013).

    CAS  PubMed  Google Scholar 

  65. 65.

    Greenbaum, S. et al. Spatio-temporal coordination at the maternal-fetal interface promotes trophoblast invasion and vascular remodeling in the first half of human pregnancy. Preprint at bioRxiv (2021).

  66. 66.

    Garrido-Gomez, T. et al. Defective decidualization during and after severe preeclampsia reveals a possible maternal contribution to the etiology. Proc. Natl Acad. Sci. USA 114, E8468–E8477 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  67. 67.

    Deep Cell Core Library. Deep learning for single-cell analysis.

  68. 68.

    Bankhead, P. et al. QuPath: open source software for digital pathology image analysis. Sci. Rep. 7, 16878 (2017).

    PubMed  PubMed Central  Google Scholar 

  69. 69.

    Graham, S. et al. Hover-Net: simultaneous segmentation and classification of nuclei in multi-tissue histology images. Med. Image Anal. 58, 101563 (2019).

    PubMed  Google Scholar 

  70. 70.

    Tsai, H.-F., Gajda, J., Sloan, T. F. W., Rares, A. & Shen, A. Q. Usiigaci: instance-aware cell tracking in stain-free phase contrast microscopy enabled by machine learning. SoftwareX 9, 230–237 (2019).

    Google Scholar 

  71. 71.

    Kiemen, A. et al. In situ characterization of the 3D microanatomy of the pancreas and pancreatic cancer at single cell resolution. bioRxiv 2020.12.08.416909 (2020)

  72. 72.

    Cao, J. et al. Establishment of a morphological atlas of the Caenorhabditis elegans embryo using deep-learning-based 4D segmentation. Nat. Commun. 11, 6254 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  73. 73.

    Schulz, D. et al. Simultaneous multiplexed imaging of mRNA and proteins with subcellular resolution in breast cancer tissue samples by mass cytometry. Cell Syst. 6, 531 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  74. 74.

    McKinley, E. T. et al. Optimized multiplex immunofluorescence single-cell analysis reveals tuft cell heterogeneity. JCI Insight 2, e93487 (2017).

    PubMed Central  Google Scholar 

  75. 75.

    Patel, S. S. et al. The microenvironmental niche in classic Hodgkin lymphoma is enriched for CTLA-4- positive T-cells that are PD-1-negative. Blood 134, 2059–2069 (2019).

    PubMed  PubMed Central  Google Scholar 

  76. 76.

    Jackson, H. W. et al. The single-cell pathology landscape of breast cancer. Nature 578, 615–620 (2020).

    CAS  PubMed  Google Scholar 

  77. 77.

    Rashid, R. et al. Highly multiplexed immunofluorescence images and single-cell data of immune markers in tonsil and lung cancer. Sci. Data 6, 323 (2019).

    PubMed  PubMed Central  Google Scholar 

  78. 78.

    McCaffrey, E. F. et al. Multiplexed imaging of human tuberculosis granulomas uncovers immunoregulatory features conserved across tissue and blood. Preprint at bioRxiv (2020).

  79. 79.

    Walt, Svander et al. scikit-image: image processing in Python. PeerJ 2, e453 (2014).

    PubMed  PubMed Central  Google Scholar 

  80. 80.

    Kingma, D. P. & Ba, J. Adam: a method for stochastic optimization. Preprint at (2014).

  81. 81.

    Kluyver, T. et al. in Positioning and Power in Academic Publishing: Players, Agents and Agendas (eds Schmidt, B. & Loizides, F.) (IOS Press, 2016).

  82. 82.

    Chollet, F. et al. Keras. (2015).

  83. 83.

    Hunter, J. D. Matplotlib: a 2D graphics environment. Comput. Sci. Eng. 9, 90–95 (2007).

    Google Scholar 

  84. 84.

    Harris, C. R. et al. Array programming with NumPy. Nature 585, 357–362 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  85. 85.

    Reback, J. et al. pandas-dev/pandas: Pandas 1.1.3. (2020).

  86. 86.

    Pedregosa, F. et al. Scikit-learn: machine learning in Python. Preprint at (2012).

  87. 87.

    Waskom, M. et al. mwaskom/seaborn. (2020).

  88. 88.

    Abadi, M. et al. TensorFlow: large-scale machine learning on heterogeneous distributed systems. Preprint at (2016).

  89. 89.

    Hoyer, S. & Hamman, J. xarray: N-D labeled arrays and datasets in Python. J. Open Res. Softw. 5, 10 (2017).

    Google Scholar 

Download references


We thank K. Borner, L. Cai, M. Covert, A. Karpathy, S. Quake and M. Thomson for interesting discussions; D. Glass and E. McCaffrey for feedback on the manuscript; T. Vora for copy editing; R. Angoshtari, G. Barlow, B. Bodenmiller, C. Carey, R. Coffey, A. Delmastro, C. Egelston, M. Hoppe, H. Jackson, A. Jeyasekharan, S. Jiang, Y. Kim, E. McCaffrey, E. McKinley, M. Nelson, S.-B. Ng, G. Nolan, S. Patel, Y. Peng, D. Philips, R. Rashid, S. Rodig, S. Santagata, C. Schuerch, D. Schulz, Di. Simons, P. Sorger, J. Weirather and Y. Yuan for providing imaging data for TissueNet; the crowd annotators who powered our human-in-the-loop pipeline; and all patients who donated samples for this study. This work was supported by grants from the Shurl and Kay Curci Foundation, the Rita Allen Foundation, the Susan E. Riley Foundation, the Pew Heritage Trust, the Alexander and Margaret Stewart Trust, the Heritage Medical Research Institute, the Paul Allen Family Foundation through the Allen Discovery Centers at Stanford and Caltech, the Rosen Center for Bioengineering at Caltech and the Center for Environmental and Microbial Interactions at Caltech (D.V.V.). This work was also supported by 5U54CA20997105, 5DP5OD01982205, 1R01CA24063801A1, 5R01AG06827902, 5UH3CA24663303, 5R01CA22952904, 1U24CA22430901, 5R01AG05791504 and 5R01AG05628705 from NIH, W81XWH2110143 from DOD, and other funding from the Bill and Melinda Gates Foundation, Cancer Research Institute, the Parker Center for Cancer Immunotherapy and the Breast Cancer Research Foundation (M.A.). N.F.G. was supported by NCI CA246880-01 and the Stanford Graduate Fellowship. B.J.M. was supported by the Stanford Graduate Fellowship and Stanford Interdisciplinary Graduate Fellowship. T.D. was supported by the Schmidt Academy for Software Engineering at Caltech.

Author information




N.F.G., L.K., M.A. and D.V.V. conceived the project. E.M. and D.V.V. conceived the human-in-the-loop approach. L.K. and M.A. conceived the whole-cell segmentation approach. G.M., T.D., E.M., W.G. and D.V.V. developed DeepCell Label. G.M., N.F.G., E.M., I.C., W.G. and D.V.V. developed the human-in-the-loop pipeline. M.S.S., C.P., W.G. and D.V.V. developed Mesmer’s deep learning architecture. W.G., N.F.G. and D.V.V. developed model training software. C.P. and W.G. developed cloud deployment. M.S.S., S.C., W.G. and D.V.V. developed metrics software. W.G. developed plugins. N.F.G., A. Kong, A. Kagel, J.S. and O.B.-T. developed the multiplex image analysis pipeline. A. Kagel and G.M. developed the pathologist evaluation software. N.F.G., G.M. and T.H. supervised training data creation. N.F.G., C.C.F., B.J.M., K.X.L., M.F., G.C., Z.A., J.M. and S.W. performed quality control on the training data. E.S., S.G. and T.R. generated MIBI-TOF data for morphological analyses. S.C.B. helped with experimental design. N.F.G., W.G. and D.V.V. trained the models. N.F.G., W.G., G.M. and D.V.V. performed data analysis. N.F.G., G.M., M.A. and D.V.V. wrote the manuscript. M.A. and D.V.V. supervised the project. All authors provided feedback on the manuscript.

Corresponding authors

Correspondence to Michael Angelo or David Van Valen.

Ethics declarations

Competing interests

M.A. is an inventor on patent US20150287578A1. M.A. is a board member and shareholder in IonPath Inc. T.R. has previously consulted for IonPath Inc. D.V.V and E.M. have filed a provisional patent for this work. The remaining authors declare no competing interests.

Additional information

Peer review information Nature Biotechnology thanks the anonymous reviewers for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 DeepCell Label annotation workflow.

a, How multichannel images are represented and edited in DeepCell Label. b, Scalable backend for DeepCell Label that dynamically adjusts required resources based on usage, allowing concurrent annotators to work in parallel. c, Human-in-the-loop workflow diagram. Images are uploaded to the server, run through Mesmer to make predictions, and cropped to facilitate error correction. These crops are sent to the crowd to be corrected, stitched back together, run through quality control to ensure accuracy, and used to train an updated model.

Extended Data Fig. 2 Mesmer benchmarking.

a, PanopticNet architecture. Images are fed into a ResNet50 backbone coupled to a feature pyramid network. Two semantic heads produce pixel-level predictions. The first head predicts whether each pixel belongs to the interior, border, or background of a cell, while the second head predicts the center of each cell. b, Relative proportion of preprocessing, inference, and postprocessing time in PanopticNet architecture. c, Evaluation of precision, recall, and Jaccard index for Mesmer and previously published models (right) and models trained on TissueNet (left). d, Summary of TissueNet accuracy for Mesmer and selected models to facilitate future benchmarking efforts e,f Breakdown of most prevalent error types (e) and less prevalent error types (f) for Mesmer and previously published models illustrates Mesmer’s advantages over previous approaches. g, Comparison of the size distribution of prediction errors for Mesmer (left) with nuclear segmentation followed by expansion (right) shows that Mesmer’s predictions are unbiased.

Extended Data Fig. 3 TissueNet accuracy comparisons.

a, Accuracy of specialist models trained on each platform type (rows) and evaluated on data from other platform types (columns) indicates good agreement within immunofluorescence and mass spectrometry-based methods, but not across distinct methods. b, Accuracy of specialist models trained on each tissue type (rows) and evaluated on data from other tissue types (columns) demonstrates that models trained on only a single tissue type do not generalize as well to other tissue types. c, Quantification of F1 score as a function of the size of the dataset used for training. d-h, Quantification of individual error types as a function of the size of the dataset used for training. i, Representative images where Mesmer accuracy was poor, as determined by the image specific F1 score. j, Impact of image blurring on model accuracy. k, Impact of image downsampling and then upsampling on model accuracy. l, Impact of adding random noise to image on model accuracy. All scale bars are 50 μM.

Extended Data Fig. 4 3D segmentation.

Proof of principle for using Mesmer’s segmentation predictions to generate 3D segmentations. A z-stack of 3D data is fed to Mesmer, which generates separate 2D predictions for each slice. We computationally link the segmentations predictions from each slice to form 3D objects. This approach can form the basis for human-in-the-loop construction of training data for 3D models.

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Greenwald, N.F., Miller, G., Moen, E. et al. Whole-cell segmentation of tissue images with human-level performance using large-scale data annotation and deep learning. Nat Biotechnol (2021).

Download citation


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing