Abstract
A central challenge in biology is obtaining high-content, high-resolution information while analyzing tissue samples at volumes relevant to disease progression. We address this here with CODA, a method to reconstruct exceptionally large (up to multicentimeter cubed) tissues at subcellular resolution using serially sectioned hematoxylin and eosin-stained tissue sections. Here we demonstrate CODA’s ability to reconstruct three-dimensional (3D) distinct microanatomical structures in pancreas, skin, lung and liver tissues. CODA allows creation of readily quantifiable tissue volumes amenable to biological research. As a testbed, we assess the microanatomy of the human pancreas during tumorigenesis within the branching pancreatic ductal system, labeling ten distinct structures to examine heterogeneity and structural transformation during neoplastic progression. We show that pancreatic precancerous lesions develop into distinct 3D morphological phenotypes and that pancreatic cancer tends to spread far from the bulk tumor along collagen fibers that are highly aligned to the 3D curves of ductal, lobular, vascular and neural structures. Thus, CODA establishes a means to transform broadly the structural study of human diseases through exploration of exhaustively labeled 3D microarchitecture.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
Data availability
Due to the extremely large size of the digital files described, data are available upon request from the corresponding author. Source data are provided with this paper.
Code availability
Code is available on the following GitHub page: https://github.com/ashleylk/CODA.
References
Liebig, C., Ayala, G., Wilks, J. A., Berger, D. H. & Albo, D. Perineural invasion in cancer. Cancer 115, 3379–3391 (2009).
Hong, S. M. et al. Three-dimensional visualization of cleared human pancreas cancer reveals that sustained epithelial-to-mesenchymal transition is not required for venous invasion. Mod. Pathol. 33, 639–647 (2019).
Kuett, L. et al. Three-dimensional imaging mass cytometry for highly multiplexed molecular and cellular mapping of tissues and the tumor microenvironment. Nat. Cancer 3, 122–133 (2021).
Siegel, R. L., Miller, K. D., Fuchs, H. E. & Jemal, A. Cancer statistics, 2021. CA Cancer J. Clin. 71, 7–33 (2021).
Michaud, D. S. et al. Physical activity, obesity, height, and the risk of pancreatic cancer. JAMA 286, 921–929 (2001).
Hruban, R. H. et al. Why is pancreatic cancer so deadly? The pathologist’s view. J. Pathol. 248, 131–141 (2019).
Tanaka, M. et al. Meta-analysis of recurrence pattern after resection for pancreatic cancer. Br. J. Surg. 106, 1590–1601 (2019).
Zhang, J.-F. et al. Influence of perineural invasion on survival and recurrence in patients with resected pancreatic cancer. Asian Pac. J. Cancer Prev. 14, 5133–5139 (2013).
Huang, L. et al. Ductal pancreatic cancer modeling and drug screening using human pluripotent stem cell- and patient-derived tumor organoids. Nat. Med. 21, 1364–1371 (2015).
Drost, J. & Clevers, H. Organoids in cancer research. Nat. Rev. Cancer 18, 407–418 (2018).
Taniuchi, K. et al. Overexpressed P-Cadherin/CDH3 promotes motility of pancreatic cancer cells by interacting with p120ctn and activating Rho-Family GTPases. Cancer Res. 65, 3092–3099 (2005).
Plentz, R. et al. Inhibition of γ-secretase activity inhibits tumor progression in a mouse model of pancreatic ductal adenocarcinoma. Gastroenterology 136, 1741–1749.e6 (2009).
Cruz-Monserrate, Z. et al. Detection of pancreatic cancer tumours and precursor lesions by cathepsin E activity in mouse models. Gut 61, 1315–1322 (2012).
Yang, B. et al. Single-cell phenotyping within transparent intact tissue through whole-body clearing. Cell 158, 945–958 (2014).
Murakami, T. C. et al. A three-dimensional single-cell-resolution whole-brain atlas using CUBIC-X expansion microscopy and tissue clearing. Nat. Neurosci. 21, 625–637 (2018).
Susaki, E. A. et al. Versatile whole-organ/body staining and imaging based on electrolyte-gel properties of biological tissues. Nat. Commun. 11, 1982 (2020).
Zhao, S. et al. Cellular and molecular probing of intact human organs. Cell 180, 796–812.e19 (2020).
Chung, K. et al. Structural and molecular interrogation of intact biological systems. Nature 497, 332–337 (2013).
Richardson, D. S. & Lichtman, J. W. Clarifying tissue clearing. Cell 162, 246–257 (2015).
Xie, W. et al. Prostate cancer risk stratification via nondestructive 3D pathology with deep learning–assisted gland analysisprostate cancer risk stratification via 3D gland analysis. Cancer Res. 82, 334–345 (2022).
Hahn, M. et al. Mesoscopic 3D imaging of pancreatic cancer and Langerhans islets based on tissue autofluorescence. Sci. Rep. 10, 18246 (2020).
Liu, J. T. C. et al. Harnessing non-destructive 3D pathology. Nat. Biomed. Eng. 5, 203–218 (2021).
Groot, A. Ede et al. Characterization of tumor-associated macrophages in prostate cancer transgenic mouse models. Prostate 81, 629–647 (2021).
Song, Y., Treanor, D., Bulpitt, A. J. & Magee, D. R. 3D reconstruction of multiple stained histology images. J. Pathol. Inform. 4, 7 (2013).
Lotz, J. M. et al. Integration of 3D multimodal imaging data of a head and neck cancer and advanced feature recognition. Biochim. Biophys. Acta: Proteins Proteom. 1865, 946–956 (2017).
Lotz, J. et al. Zooming in: high resolution 3D reconstruction of differently stained histological whole slide images. Proc Spie 9041, 16–22 (2014).
Tempest, N. et al. Histological 3D reconstruction and in vivo lineage tracing of the human endometrium. J. Pathol. 251, 440–451 (2020).
Rees, J. et al. O36 Investigating clonal expansions in the normal stomach and the 3D architecture of oxyntic gastric glands. Gut 70, A20–A21 (2021).
Graham, S. et al. Hover-Net: simultaneous segmentation and classification of nuclei in multi-tissue histology images. Med. Image Anal. 58, 101563 (2019).
Bankhead, P. et al. QuPath: open source software for digital pathology image analysis. Sci. Rep. 7, 16878 (2017).
Chan, L. et al. HistoSegNet: semantic segmentation of histological tissue type in whole slide images, in Proc. International Conference on Computer Vision (ICCV) 2019, Seoul, Korea 10662–10671 (ICCV, 2019).
Ternes, L. et al. VISTA: visual semantic tissue analysis for pancreatic disease quantification in murine cohorts. Sci. Rep. 10, 20904 (2020).
Magee, D. et al. Histopathology in 3D: from three-dimensional reconstruction to multi-stain and multi-modal analysis. J. Pathol. Inform. 6, 6 (2015).
Roberts, N. et al. Toward routine use of 3D histopathology as a research tool. Am. J. Pathol. 180, 1835–1842 (2012).
Kartasalo, K. et al. Comparative analysis of tissue reconstruction algorithms for 3D histology. Bioinformatics 34, 3013 (2018).
Wu, P. H. et al. High-throughput ballistic injection nanorheology to measure cell mechanics. Nat. Protoc. 7, 155–170 (2012).
Chen, L.-C., Zhu, Y., Papandreou, G., Schroff, F. & Adam, H. Encoder-decoder with atrous separable convolution for semantic image segmentation, in Proc. European Conference on Computer Vision, ECCV 2018 (eds Ferrari, V. et al.) 883–851 (Springer, 2018).
Yoshizawa, T. et al. Three-dimensional analysis of extrahepatic cholangiocarcinoma and tumor budding. J. Pathol. 251, 400–410 (2020).
Basturk, O. et al. A revised classification system and recommendations from the baltimore consensus meeting for neoplastic precursor lesions in the pancreas. Am. J. Surg. Pathol. 39, 1730 (2015).
Singhi, A. D., Koay, E. J., Chari, S. T. & Maitra, A. Early detection of pancreatic cancer: opportunities and challenges. Gastroenterology 156, 2024–2040 (2019).
Hruban, R. H., Maitra, A. & Goggins, M. Update on pancreatic intraepithelial neoplasia. Int. J. Clin. Exp. Pathol. 1, 306 (2008).
Canto, M. I. et al. Screening for early pancreatic neoplasia in high-risk individuals: a prospective controlled study. Clin. Gastroenterol. Hepatol. 4, 766–781 (2006).
Zhu, L., Shi, G., Schmidt, C. M., Hruban, R. H. & Konieczny, S. F. Acinar cells contribute to the molecular heterogeneity of pancreatic intraepithelial neoplasia. Am. J. Pathol. 171, 263–273 (2007).
Morris, J. P. IV, Cano, D. A., Sekine, S., Wang, S. C. & Hebrok, M. β-catenin blocks Kras-dependent reprogramming of acini into pancreatic cancer precursor lesions in mice. J. Clin. Invest. 120, 508–520 (2010).
Messal, H. A. et al. Tissue curvature and apicobasal mechanical tension imbalance instruct cancer morphogenesis. Nature 566, 126–130 (2019).
Xu, S. et al. The role of collagen in cancer: from bench to bedside. J. Transl. Med. 17, 309 (2019).
Puls, T. J., Tan, X., Whittington, C. F. & Voytik-Harbin, S. L. 3D collagen fibrillar microstructure guides pancreatic cancer cell phenotype and serves as a critical design parameter for phenotypic models of EMT. PLoS ONE 12, e0188870 (2017).
Drifka, C. R. et al. Highly aligned stromal collagen is a negative prognostic factor following pancreatic ductal adenocarcinoma resection. Oncotarget 7, 76197 (2016).
Drifka, C. R. et al. Periductal stromal collagen topology of pancreatic ductal adenocarcinoma differs from that of normal and chronic pancreatitis. Mod. Pathol. 28, 1470–1480 (2015).
Sunderland, S. S. The anatomy and physiology of nerve injury. Muscle Nerve 13, 771–784 (1990).
Lundborg, G. & Dahlin, L. B. Anatomy, function, and pathophysiology of peripheral nerves and nerve compression. Hand Clin. 12, 185–193 (1996).
Axer, H., Axerl, M., Krings, T. & Keyserlingk, D. G. V. Quantitative estimation of 3-D fiber course in gross histological sections of the human brain using polarized light. J. Neurosci. Methods 105, 121–131 (2001).
Fraley, S. I. et al. Three-dimensional matrix fiber alignment modulates cell migration and MT1-MMP utility by spatially and temporally directing protrusions. Sci. Rep. 5, 14580 (2015).
Rios, A. C. et al. Intraclonal plasticity in mammary tumors revealed through large-scale single-cell resolution 3D imaging. Cancer Cell 35, 618–632.e6 (2019).
Cuccarese, M. F. et al. Heterogeneity of macrophage infiltration and therapeutic response in lung carcinoma revealed by 3D organ imaging. Nat. Commun. 8, 14293 (2017).
Lai, H. M. et al. Antibody stabilization for thermally accelerated deep immunostaining. Nat. Methods https://doi.org/10.1038/s41592-022-01569-1 (2022).
Saltz, J. et al. Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images. Cell Rep. 23, 181–193.e7 (2018).
Lehmann, B. D. et al. Refinement of triple-negative breast cancer molecular subtypes: implications for neoadjuvant chemotherapy selection. PLoS ONE 11, e0157368 (2016).
Nirschl, J. J. et al. A deep-learning classifier identifies patients with clinical heart failure using whole-slide images of H&E tissue. PLoS ONE 13, e0192726 (2018).
Goode, A., Gilbert, B., Harkes, J., Jukic, D. & Satyanarayanan, M. OpenSlide: a vendor-neutral software foundation for digital pathology. J. Pathol. Inform. 4, 27 (2013).
Falkena, W. xml2struct v.1.8.0.0 (MathWorks, 2020).
Hoffmann, H. Simple violin plot using matlab default kernel density estimation (INRES, Univ. Bonn, 2015).
Acknowledgements
We thank J. Phillip and D. Gilkes for their important feedback in this work. We would additionally like to thank sources of funding for additional projects in our groups: grant nos. NIH/NCI P50 CA62924; NIH/NIDDK K08 DK107781; Sol Goldman Pancreatic Cancer Research Center; Buffone Family Gastrointestinal Cancer Research Fund; Carol S. and Robert M. Long Pancreatic Cancer Research Fund; Allegheny Health Network, Johns Hopkins Cancer Research Fund; American Cancer Society, The Cornelia T. Bailey Foundation Research Scholar grant no. RSG-18-143-01; AACR-Bristol-Myers Squibb Midcareer Female Investigator Grant; Emerson Collective Cancer Research Fund; Robert L. Fine Pancreatic Cancer Research Foundation; Rolfe Pancreatic Cancer Foundation; Joseph C Monastra Foundation; The Gerald O Mann Charitable Foundation (H. and A. Wulfstat, Trustees); S. Wojcicki and D. Troper; The Carl and Carol Nale Fund for Pancreatic Cancer Research. The Johns Hopkins University Oncology Tissue Services core used for sectioning and staining is funded by the SKCCC Cancer Center Support grant (CCSG; grant no. P30 CA006973). Funding came from the National Institutes of Health/National Cancer Institute grant no. U54CA268083. (D.W., P.W. and A.L.K.); National Institutes of Health/National Cancer Institute grant no. U54CA210173. (D.W.); National Institutes of Health/National Institute on Aging grant no. U01AG060903 (D.W.); National Institutes of Health/National Cancer Institute grant no. UG3CA275681 (P.W.); The Sol Goldman Pancreatic Cancer Research Center (A.M.B., L.D.W., F.A., E.D.T., R.H.H., P.W. and D.W.); S. Wojcicki and D. Troper (A.L.K., A.M.B., L.D.W., F.A. and E.D.T.); The Rolfe Foundation for Pancreatic Cancer Research, Allegheny Health Network—Johns Hopkins Cancer Research Fund (A.M.B.); ARCS Foundation, Inc. (A.L.K.); Nanotechnology for Cancer Research T32 Training grant no. 5T32CA153952 (A.L.K.) and an NVIDIA GPU grant (P.W.).
Author information
Authors and Affiliations
Contributions
Conceptualization was done by L.D.W., R.H.H., P.-H.W. and D.W. Image registration methodology was carried out by P.-H.W. and A.L.K. Deep learning methodology was done by A.L.K. 3D reconstruction methodology was done by A.L.K. 3D quantification methodology was devised by P.-H.W. and A.L.K. CODA validation was done by A.C.J., P.-H.W. and A.L.K. Collagen alignment methodology was developed by P.-H.W., K.S.H. and A.L.K. Tissue annotation was done by M.P.G., A.M.B., J.M.B., R.R., F.A., A.L.K., A.C.J., B.K. and J.H. Tissue collection, sectioning and scanning were carried out by S.R., T.C.C. and A.M.B. Pathology consultation was done by S.-M.H., E.D.T., L.D.W. and R.H.H. Biostatistics calculations were done by A.L.K., P.-H.W. and P.H. The original draft was written by A.L.K., P.-H.W. and D.W. Review and editing of the draft were done by D.W., P.-H.W., L.D.W., R.H.H., P.H., M.P.G., A.M.B., J.M.B., R.R., F.A., A.C.J., B.K., J.H., K.S.H., S.M.H., E.D.T., T.C.C., S.R. and A.L.K.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Primary Handling Editor: Madhura Mukhopadhyay, in collaboration with the Nature Methods team.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Histological image registration sample workflow.
Tissue cases registered with reference at center z-height of sample. Example fixed and moving images shown. Global registration performed with rotational reference at center of fixed image. Fixed and moving images smoothed by conversion to greyscale, removal of non-tissue objects in image, intensity complementing, and Gaussian filtering to reduce pixel-level noise in images. Radon transforms calculated filtered fixed and moving for discrete degrees 0–360. Maximum of 2D cross correlation of radon transforms yields registration angle. Maximum of 2D cross correlation of filtered images yields registration translation. Local registration performed at discrete intervals across fixed image. For each reference point, tiles are cropped from fixed and moving images and coarse registration is performed on tiles. Results from all tiles are interpolated on 2D grids to create nonlinear whole-image displacement fields. Sample overlays of pre and postregistration.
Extended Data Fig. 2 Overview of semantic segmentation workflow and training data design.
(a) For each case, a minimum of seven images were extracted from for manual annotation. For each extracted image, minimum 50 examples of each tissue type were annotated, and the annotations cropped from the larger image. (b) To construct training and validation sets, cropped annotations were overlayed on a large image until the image was >65% full and such that the number of annotations of each type was roughly equal. (c) These large tiles were cut into smaller tiles for training and validation. Additional tiles were created for the testing set where the annotation was not cropped from the image. Testing accuracy was assessed as the percentage of the annotated area of the tile classified correctly. Following model training, the serial images were cropped into tiles and semantically segmented.
Extended Data Fig. 3 Additional methodological supplement.
(a) Sample predicted vs. true outcomes for deep learning models for sample P1 (left) and P8 (right). (b) Workflow for creation of multi-patient semantic segmentation of nerves. Nerve annotations collected from thirteen pancreas samples. Original tissue annotations reformatted to: 1. smooth muscle, 2. collagen, 3. other tissue (islets, normal ducts, acini, precursor, lymph, PDAC), 4. white (whitespace, fat). Nerve annotations combined with original annotations to create a dataset for nerve recognition in H&E images. (c) Sample P7 average and per class testing accuracy as a function of percent of training annotations used. (d) Incidence of pancreatic phenotypes in eight samples. (e) Comparison of nuclear aspect ratio measurements performed by person 1 and person 2 (N = 150 nuclei per person per condition) show nonsignificant differences between measurements using the Wilcoxon rank sum test.
Extended Data Fig. 4 Validation of cell count and 2D to 3D cell count extrapolation.
(a) Sample histological section and corresponding color deconvolved hematoxylin channel of image. All cells in five validation images were manually annotated by two persons. Annotations were compared to CODA outputs and outputs from two existing cell counting methods27,28. (b) Cell diameters of each tissue subtype were measured using Aperio ImageScope. 2D cell counts were extrapolated to 3D using the formula listed. It was assumed that cells could be detected by the algorithm if any part of the nucleus touched the tissue section. Therefore, effective tissue section thickness equals true tissue section thickness plus the diameter of the cell. 3D cell counts were estimated by multiplying 2D cell counts by the true thickness of the tissue section, multiplied by three because two sections were skipped during scanning, divided by the effective thickness of the section.
Extended Data Fig. 5 Sample histology of venous invasions identified in samples.
Thirteen distinct venous invasions were identified in eight of the thirteen samples. For each, an H&E image was reviewed to confirm the venous invasion.
Extended Data Fig. 6 Sample histology of perineural / neural invasions identified in samples.
Ten distinct neural invasions were identified in seven of the thirteen samples, many containing regions of perineural invasion. For each, an H&E image was reviewed to confirm the structure.
Extended Data Fig. 7 Sample histology of invasion along regions of aligned collagen.
Nine distinct regions of invasion along aligned collagen were identified in five of the thirteen samples, including invasion along periductal collagen, invasion along perivascular collagen, and invasion along interlobular collagen. For each, an H&E image was reviewed to confirm the structure.
Supplementary information
Supplementary Information
Supplementary Tables 1–3.
Supplementary Video 1
Video 1. 3D reconstruction of P1: normal human pancreas.
Supplementary Video 2
Video 2. 3D reconstruction of P2: human pancreas containing PanIN.
Supplementary Video 3
Video 3. 3D reconstruction of P5: human pancreas containing IPMN.
Supplementary Video 4
Video 4. 3D reconstruction of P7: human pancreas containing PDAC.
Supplementary Video 5
Video 5. 3D reconstruction of P8: human pancreas containing PDAC.
Supplementary Video 6
Video 6. 3D reconstruction of P11: human pancreas containing PDAC.
Supplementary Video 7
Video 7. Identification of 38 distinct PanIN in a sample of human pancreas in P2.
Supplementary Video 8
Video 8. Identification of two phenotypes of pancreatic precancer in P2.
Source data
Source Data Fig. 1
Data for Fig. 2a–d registration methods comparison raw data, normalized registration performance graph, change in tissue composition with skipping sections and change in cell count with skipping sections.
Source Data Fig. 4
Data for Fig. 4b normal versus cancer cell density, and data for Fig. 4e comparison of participants 1 and 2 nuclear aspect ratio measurements.
Source Data Fig. 5
Data for Fig. 5b precursor 2D to 3D count overestimation factor.
Source Data Fig. 6
Data for Fig. 6c extracellular matrix anisotropy index and nuclear aspect ratio.
Source Data Extended Data Fig. 2
Data for Extended Data Fig. 4a precision and recall data.
Source Data Extended Data Fig. 4
Data for Extended Data Fig. 2c per-class testing accuracy as a function of percentage of annotations used.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Kiemen, A.L., Braxton, A.M., Grahn, M.P. et al. CODA: quantitative 3D reconstruction of large tissues at cellular resolution. Nat Methods 19, 1490–1499 (2022). https://doi.org/10.1038/s41592-022-01650-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41592-022-01650-9
This article is cited by
-
Exploring chronic and transient tumor hypoxia for predicting the efficacy of hypoxia-activated pro-drugs
npj Systems Biology and Applications (2024)
-
Intraoperative margin assessment for basal cell carcinoma with deep learning and histologic tumor mapping to surgical site
npj Precision Oncology (2024)
-
Spatial landmark detection and tissue registration with deep learning
Nature Methods (2024)
-
Mapping cell-to-tissue graphs across human placenta histology whole slide images using deep learning with HAPPY
Nature Communications (2024)
-
Virtual alignment of pathology image series for multi-gigapixel whole slide images
Nature Communications (2023)