Search

[This corrects the article DOI: 10.1371/journal.pcbi.1008412.].

Background: Automated cell-level characterization of the tumor microenvironment (TME) at scale is key to data-driven immuno-oncology. Artificial intelligence (AI)-powered analysis of hematoxylin and eosin (H&E) images scales and has recently been translated into diagnostics. However, robust TME analysis solely based on H&E data is bound by the stain's properties and by manual pathologist annotations, both in number and accuracy. In this study, we quantify the error introduced by pathologists' morphological assessment and mitigate this error by training AI-systems without manual pathologist annotations, using labels determined directly from IHC profiles. Methods: The work was carried out on 239 clinical NSCLC cases. CK-KL1, CD3+CD20, and Mum1 were used for defining carcinoma (CA), lymphocyte (LY), and plasma (PL) cells. For evaluation, representative regions were annotated by 3 trained pathologists. The workflow is based on co-registration of same-section H&E and IHC stained images with single cell precision. Cells were detected in H&E and labelled using rule-based algorithms that incorporated IHC information. This H&E data was used to train neural networks (NN). Results: (A) The inter-rater agreement of pathologists annotating on H&E is increased when information from registered IHC images is provided. (B) The concordance of pathologists on H&E-only compared to on H&E+IHC shows that pathologists miss or misclassify cells with a certain error. (C) NNs trained with IHC-based labels achieve similar performance for cell type classification on H&E as pathologists on H&E. Conclusion: This study demonstrates the value of combining histomorphological and IHC data for improved cell annotation. Our novel workflow provides a quantitative benchmark and facilitates training of accurate AI models for quantitative characterization of tumor and TME from H&E sections. A) Inter-rater agreement by metric, stain, and cell type By cell count, Pearson correlation By cell count, Pearson correlation By cell location, Krippendorff’s alpha By cell location, Krippendorff’s alpha Cell type H&E-only H&E+IHC H&E-only H&E+IHC CA 0.86 0.98 0.43 0.90 LY 0.88 0.99 0.21 0.76 PL 0.77 0.96 0.32 0.87 B) Performance of individual pathologists in H&E Against consensus in H&E+IHC Against own annotations in H&E+IHC Against own annotations in H&E+IHC Cell type By cell count, Pearson correlation By cell location, Precision By cell location, Recall CA 0.84 0.76 0.77 LY 0.78 0.70 0.60 PL 0.76 0.69 0.21 C) NN against annotator H&E+IHC consensus Cell Type By cell count, Pearson correlation CA 0.84 LY 0.92 PL 0.75 Citation Format: Thomas Mrowiec, Sharon Ruane, Simon Schallenberg, Gabriel Dernbach, Rumyana Todorova, Cornelius Böhm, Walter de Back, Blanca Pablos, Roman Schulte-Sasse, Ivana Trajanovska, Adelaida Creosteanu, Emil Barbuta, Marcus Otte, Christian Ihling, Hans Juergen Grote, Juergen Scheuenpflug, Viktor Matyas, Maximilian Alber, Frederick Klauschen. Immunohistochemistry-informed AI systems for improved characterization of tumor-microenvironment in clinical non-small cell lung cancer H&E samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 457.

How epithelial cells coordinate their polarity to form functional tissues is an open question in cell biology. Here, we characterize a unique type of polarity found in liver tissue, nematic cell polarity, which is different from vectorial cell polarity in simple, sheet-like epithelia. We propose a conceptual and algorithmic framework to characterize complex patterns of polarity proteins on the surface of a cell in terms of a multipole expansion. To rigorously quantify previously observed tissue-level patterns of nematic cell polarity (Morales-Navarrete et al., eLife 2019), we introduce the concept of co-orientational order parameters, which generalize the known biaxial order parameters of the theory of liquid crystals. Applying these concepts to three-dimensional reconstructions of single cells from high-resolution imaging data of mouse liver tissue, we show that the axes of nematic cell polarity of hepatocytes exhibit local coordination and are aligned with the biaxially anisotropic sinusoidal network for blood transport. Our study characterizes liver tissue as a biological example of a biaxial liquid crystal. The general methodology developed here could be applied to other tissues and in-vitro organoids., Author summary Cell polarity enables cells to carry out specific functions. Cell polarity is characterized by the formation of different plasma membrane domains, each with specific composition of proteins, phospholipids and cytoskeletal components. In simple epithelial sheets, or tube-like tissues such as kidney, epithelial cells are known to display a single apical domain, facing a lumenal cavity, and a single basal domain on the opposite side of the cell, facing a basal layer of extracellular matrix. This apico-basal polarity defines a vector of cell polarity, which provides a direction of fluid transport, e.g., from the basal side of the sheet to the lumen-facing side. In more complex, three-dimensional epithelial tissues, such as liver tissue with its complex network of blood-transporting sinusoids, the membrane domains of hepatocyte cells display more intricate patterns, including rings and antipodal pairs of apical membrane. Here, we develop a mathematical framework to precisely characterize and quantify complex polarity patterns. Thereby, we reveal ordered patterns of cell polarity that span across a liver lobule. Our new method builds on physical concepts originally developed for ordered phases of liquid crystals. It provides a versatile tool to characterize the spatial organization of a complex three-dimensional tissue.

The prediction of streamflow from precipitation data is one of the traditional disciplines of hydrological modelling and has major societal implications such as flood forecasting, efficient use of hydro-power and urban and regional planning. Recently, data-driven approaches have been applied successfully for rainfall-runoff modelling, often outperforming equivalent physical modeling approaches. However, these studies have almost exclusively focused on temporal data and have neglected data on the spatial distribution of the inputs.To close this gap, we trained convolutional long-short-term-memory (ConvLSTM) models on daily temperature and precipitation maps of the catchment area to predict the streamflow of the Elbe river. This supervised deep learning method combines convolutional and recurrent neural networks to extract useful features in the spatio-temporal input maps to predict the river’s streamflow. We embedded the model into a Bayesian framework to deliver estimates of prediction uncertainty along with the predictions. Moreover, we derived saliency maps that highlight the most relevant patterns in precipitation and temperature for the Elbe‘s major flood events.Comparison with physical simulations show that our Bayesian ConvLSTM approach (1) performs on par with results from physical modeling while requiring only input data on temperature and precipitation, (2) provides useful uncertainty estimates, and (3) is able to generate interpretable saliency maps of flooding events.In conclusion, this study showcases the applicability of deep learning methods for rainfall-runoff modelling as well as the methods' potential to gain spatial insight into the hydrological system.

Statistical and mathematical modeling are crucial to describe, interpret, compare, and predict the behavior of complex biological systems including the organization of hematopoietic stem and progenitor cells in the bone marrow environment. The current prominence of high-resolution and live-cell imaging data provides an unprecedented opportunity to study the spatiotemporal dynamics of these cells within their stem cell niche and learn more about aberrant, but also unperturbed, normal hematopoiesis. However, this requires careful quantitative statistical analysis of the spatial and temporal behavior of cells and the interaction with their microenvironment. Moreover, such quantification is a prerequisite for the construction of hypothesis-driven mathematical models that can provide mechanistic explanations by generating spatiotemporal dynamics that can be directly compared to experimental observations. Here, we provide a brief overview of statistical methods in analyzing spatial distribution of cells, cell motility, cell shapes, and cellular genealogies. We also describe cell-based modeling formalisms that allow researchers to simulate emergent behavior in a multicellular system based on a set of hypothesized mechanisms. Together, these methods provide a quantitative workflow for the analytic and synthetic study of the spatiotemporal behavior of hematopoietic stem and progenitor cells.

With ageing, intrinsic haematopoietic stem cell (HSC) activity decreases, resulting in impaired tissue homeostasis, reduced engraftment following transplantation and increased susceptibility to diseases. However, whether ageing also affects the HSC niche, and thereby impairs its capacity to support HSC function, is still widely debated. Here, by using in-vivo long-term label-retention assays we demonstrate that aged label-retaining HSCs, which are, in old mice, the most quiescent HSC subpopulation with the highest regenerative capacity and cellular polarity, reside predominantly in perisinusoidal niches. Furthermore, we demonstrate that sinusoidal niches are uniquely preserved in shape, morphology and number on ageing. Finally, we show that myeloablative chemotherapy can selectively disrupt aged sinusoidal niches in the long term, which is linked to the lack of recovery of endothelial Jag2 at sinusoids. Overall, our data characterize the functional alterations of the aged HSC niche and unveil that perisinusoidal niches are uniquely preserved and thereby protect HSCs from ageing.

Statistical and mathematical modeling are crucial to describe, interpret, compare, and predict the behavior of complex biological systems including the organization of hematopoietic stem and progenitor cells in the bone marrow environment. The current prominence of high-resolution and live-cell imaging data provides an unprecedented opportunity to study the spatiotemporal dynamics of these cells within their stem cell niche and learn more about aberrant, but also unperturbed, normal hematopoiesis. However, this requires careful quantitative statistical analysis of the spatial and temporal behavior of cells and the interaction with their microenvironment. Moreover, such quantification is a prerequisite for the construction of hypothesis-driven mathematical models that can provide mechanistic explanations by generating spatiotemporal dynamics that can be directly compared to experimental observations. Here, we provide a brief overview of statistical methods in analyzing spatial distribution of cells, cell motility, cell shapes, and cellular genealogies. We also describe cell-based modeling formalisms that allow researchers to simulate emergent behavior in a multicellular system based on a set of hypothesized mechanisms. Together, these methods provide a quantitative workflow for the analytic and synthetic study of the spatiotemporal behavior of hematopoietic stem and progenitor cells.

Functional tissue architecture originates by self-assembly of distinct cell types, following tissue-specific rules of cell-cell interactions. In the liver, a structural model of the lobule was pioneered by Elias in 1949. This model, however, is in contrast with the apparent random 3D arrangement of hepatocytes. Since then, no significant progress has been made to derive the organizing principles of liver tissue. To solve this outstanding problem, we computationally reconstructed 3D tissue geometry from microscopy images and analyzed it applying soft-condensed-matter-physics concepts. Surprisingly, analysis of the spatial organization of cell polarity revealed that hepatocytes are not randomly oriented but follow a long-range liquid-crystal order. This does not depend exclusively on hepatocytes receiving instructive signals by endothelial cells as generally assumed, since silencing Integrin-ß1 disrupted both liquid-crystal order and organization of the sinusoidal network. Our results suggest that bi-directional communication between hepatocytes and sinusoids underlies the self-organization of liver tissue.

The human epidermal growth factor receptor 2 (HER2) gene amplification status is a crucial marker for evaluating clinical therapies of breast or gastric cancer. We propose a deep learning-based pipeline for the detection, localization and classification of interphase nuclei depending on their HER2 gene amplification state in Fluorescence in situ hybridization (FISH) images. Our pipeline combines two RetinaNet-based object localization networks which are trained (1) to detect and classify interphase nuclei into distinct classes normal, low-grade and high-grade and (2) to detect and classify FISH signals into distinct classes HER2 or centromere of chromosome 17 (CEN17). By independently classifying each nucleus twice, the two-step pipeline provides both robustness and interpretability for the automated detection of the HER2 amplification status. The accuracy of our deep learning-based pipeline is on par with that of three pathologists and FISH images on a set of 57 validation images containing several hundreds of nuclei are accurately classified. The automatic pipeline is a first step towards assisting pathologists in evaluating the HER2 status of tumors using FISH images, for analyzing FISH images in retrospective studies, and for optimizing the documentation of each tumor sample by automatically annotating and reporting of the HER2 gene amplification specificities.

Pluripotent embryonic stem cells (ESCs) have the potential to differentiate into cells of all three germ layers. This unique property has been extensively studied on the intracellular, transcriptional level. However, ESCs typically form clusters of cells with distinct size and shape, and establish spatial structures that are vital for the maintenance of pluripotency. Even though it is recognized that the cells' arrangement and local interactions play a role in fate decision processes, the relations between transcriptional and spatial patterns have not yet been studied. We present a systems biology approach which combines live-cell imaging, quantitative image analysis, and multiscale, mathematical modeling of ESC growth. In particular, we develop quantitative measures of the morphology and of the spatial clustering of ESCs with different expression levels and apply them to images of both in vitro and in silico cultures. Using the same measures, we are able to compare model scenarios with different assumptions on cell–cell adhesions and intercellular feedback mechanisms directly with experimental data. Applying our methodology to microscopy images of cultured ESCs, we demonstrate that the emerging colonies are highly variable regarding both morphological and spatial fluorescence patterns. Moreover, we can show that most ESC colonies contain only one cluster of cells with high self-renewing capacity. These cells are preferentially located in the interior of a colony structure. The integrated approach combining image analysis with mathematical modeling allows us to reveal potential transcription factor related cellular and intercellular mechanisms behind the emergence of observed patterns that cannot be derived from images directly. © 2015 International Society for Advancement of Cytometry

The cell fate decision of multi-potent pancreatic progenitor cells between the exocrine and endocrine lineages is regulated by Notch signalling, mediated by cell–cell interactions. However, canonical models of Notch-mediated lateral inhibition cannot explain the scattered spatial distribution of endocrine cells and the cell-type ratio in the developing pancreas. Based on evidence from acinar-to-islet cell transdifferentiationin vitro, we propose that lateral stabilization, i.e. positive feedback between adjacent progenitor cells, acts in parallel with lateral inhibition to regulate pattern formation in the pancreas. A simple mathematical model of transcriptional regulation and cell–cell interaction reveals the existence of multi-stability of spatial patterns whose simultaneous occurrence causes scattering of endocrine cells in the presence of noise. The scattering pattern allows for control of the endocrine-to-exocrine cell-type ratio by modulation of lateral stabilization strength. These theoretical results suggest a previously unrecognized role for lateral stabilization in lineage specification, spatial patterning and cell-type ratio control in organ development.

Background Replacement of dysfunctional β-cells in the islets of Langerhans by transdifferentiation of pancreatic acinar cells has been proposed as a regenerative therapy for diabetes. Adult acinar cells spontaneously revert to a multipotent state upon tissue dissociation in vitro and can be stimulated to redifferentiate into β-cells. Despite accumulating evidence that contact-mediated signals are involved, the mechanisms regulating acinar-to-islet cell transdifferentiation remain poorly understood. Results In this study, we propose that the crosstalk between two contact-mediated signaling mechanisms, lateral inhibition and lateral stabilization, controls cell fate stability and transdifferentiation of pancreatic cells. Analysis of a mathematical model combining gene regulation with contact-mediated signaling reveals the multistability of acinar and islet cell fates. Inhibition of one or both modes of signaling results in transdifferentiation from the acinar to the islet cell fate, either by dedifferentiation to a multipotent state or by direct lineage switching. Conclusions This study provides a theoretical framework to understand the role of contact-mediated signaling in pancreatic cell fate control that may help to improve acinar-to-islet cell transdifferentiation strategies for β-cell neogenesis.

During embryonic vasculogenesis, endothelial precursor cells of mesodermal origin known as angioblasts assemble into a characteristic network pattern. Although a considerable amount of markers and signals involved in this process have been identified, the mechanisms underlying the coalescence of angioblasts into this reticular pattern remain unclear. Various recent studies hypothesize that autocrine regulation of the chemoattractant vascular endothelial growth factor (VEGF) is responsible for the formation of vascular networks in vitro. However, the autocrine regulation hypothesis does not fit well with reported data on in vivo early vascular development. In this study, we propose a mathematical model based on the alternative assumption that endodermal VEGF signalling activity, having a paracrine effect on adjacent angioblasts, is mediated by its binding to the extracellular matrix (ECM). Detailed morphometric analysis of simulated networks and images obtained from in vivo quail embryos reveals the model mimics the vascular patterns with high accuracy. These results show that paracrine signalling can result in the formation of fine-grained cellular networks when mediated by angioblast-produced ECM. This lends additional support to the theory that patterning during early vascular development in the vertebrate embryo is regulated by paracrine signalling.

Food webs are complex networks of trophic interactions in ecological communities that are crucial in creating and maintaining biodiversity, and are prominent examples of biological complexity. In this paper, we present an individual-based model of an artificial ecology demonstrating the emergence of complex food webs through the evolution of rich phenotypes. Individuals are simple structures that map several traits in a nonlinear fashion. Interaction and evolution of these structures leads to the self-assembly of food webs in complex ecological communities. Ecological and network analysis of the evolved artificial ecologies shows remarkable similarities in various patterns known from natural ecological communities.

Agent Based Modeling (ABM) is a popular technique for dealing with complex systems. An ABM usually consists of many autonomous, interacting agents, and modelers are interested in the system-level, emergent behavior of these agents. In developing an ABM, scalability is one of most critical factors for validation. Looking for an acceptable solution, parallelization often comes into play. However, writing a parallel version of an ABM simulation is at least as hard as developing the original model, and usually takes an expert of the area. This paper demonstrates our ongoing developments based on the idea that ABMs can be classified on the basis of their interior communication topology. We have developed six reusable parallel simulation schemas that can be instantiated with simulation-specific code using the Java language. Our aim was to give general, domain independent support for ABM modelers, where the parallel piece of code is completely transparent. The hope is that ABM modelers can treat their parallel system in almost the same way as they do the original. The paper details our approach as well as the implementation and, towards the end, shows performance results and how one of the templates works in a GRID system.

Summary: Morpheus is a modeling environment for the simulation and integration of cell-based models with ordinary differential equations and reaction-diffusion systems. It allows rapid development of multiscale models in biological terms and mathematical expressions rather than programming code. Its graphical user interface supports the entire workflow from model construction and simulation to visualization, archiving and batch processing. Availability and implementation: Binary packages are available at http://imc.zih.tu-dresden.de/wiki/morpheus for Linux, Mac OSX and MS Windows. Contact: walter.deback@tu-dresden.de Supplementary information: Supplementary data are available at Bioinformatics online.