Graph-Based Disease Prediction in Neuroimaging: Investigating the Impact of Feature Selection.

In biomedical machine learning, data often appear in the form of graphs. Biological systems such as protein interactions and ecological or brain networks are instances of applications that benefit from graph representations. Geometric deep learning is an arising field of techniques that has extended deep neural networks to non-Euclidean domains such as graphs. In particular, graph convolutional neural networks have achieved advanced performance in semi-supervised learning in those domains. Over the last years, these methods have gained traction in neuroscience as they could be the key to a deeper understanding in clinical diagnosis at the systems or network level (for an individual brain but also for across a cohort of subjects). As a proof-of-principle, we study and validate a previous implementation of graph-based semi-supervised classification using a ridge classifier and graph convolutional neural networks. The models are trained on population graphs that integrate imaging and phenotypic information. Our analysis employs neuroimaging data of structural and functional connectivity for prediction of neurodevelopmental and neurodegenerative disorders. Here, we particularly study the effect of different strategies to reduce the dimensionality of the neuroimaging features on the graph nodes on the classification performance.

linus: Conveniently explore, share, and present large-scale biological trajectory data in a web browser.

In biology, we are often confronted with information-rich, large-scale trajectory data, but exploring and communicating patterns in such data can be a cumbersome task. Ideally, the data should be wrapped with an interactive visualisation in one concise packet that makes it straightforward to create and test hypotheses collaboratively. To address these challenges, we have developed a tool, linus, which makes the process of exploring and sharing 3D trajectories as easy as browsing a website. We provide a python script that reads trajectory data, enriches them with additional features such as edge bundling or custom axes, and generates an interactive web-based visualisation that can be shared online. linus facilitates the collaborative discovery of patterns in complex trajectory data.

Involvement of GPR17 in Neuronal Fibre Outgrowth.

Characterization of new pharmacological targets is a promising approach in research of neurorepair mechanisms. The G protein-coupled receptor 17 (GPR17) has recently been proposed as an interesting pharmacological target, e.g., in neuroregenerative processes. Using the well-established ex vivo model of organotypic slice co-cultures of the mesocortical dopaminergic system (prefrontal cortex (PFC) and substantia nigra/ventral tegmental area (SN/VTA) complex), the influence of GPR17 ligands on neurite outgrowth from SN/VTA to the PFC was investigated. The growth-promoting effects of Montelukast (MTK; GPR17- and cysteinyl-leukotriene receptor antagonist), the glial cell line-derived neurotrophic factor (GDNF) and of two potent, selective GPR17 agonists (PSB-16484 and PSB-16282) were characterized. Treatment with MTK resulted in a significant increase in mean neurite density, comparable with the effects of GDNF. The combination of MTK and GPR17 agonist PSB-16484 significantly inhibited neuronal growth. qPCR studies revealed an MTK-induced elevated mRNA-expression of genes relevant for neuronal growth. Immunofluorescence labelling showed a marked expression of GPR17 on NG2-positive glia. Western blot and RT-qPCR analysis of untreated cultures suggest a time-dependent, injury-induced stimulation of GPR17. In conclusion, MTK was identified as a stimulator of neurite fibre outgrowth, mediating its effects through GPR17, highlighting GPR17 as an interesting therapeutic target in neuronal regeneration.

Developing 3D microscopy with CLARITY on human brain tissue: Towards a tool for informing and validating MRI-based histology.

Recent breakthroughs in magnetic resonance imaging (MRI) enabled quantitative relaxometry and diffusion-weighted imaging with sub-millimeter resolution. Combined with biophysical models of MR contrast the emerging methods promise in vivo mapping of cyto- and myelo-architectonics, i.e., in vivo histology using MRI (hMRI) in humans. The hMRI methods require histological reference data for model building and validation. This is currently provided by MRI on post mortem human brain tissue in combination with classical histology on sections. However, this well established approach is limited to qualitative 2D information, while a systematic validation of hMRI requires quantitative 3D information on macroscopic voxels. We present a promising histological method based on optical 3D imaging combined with a tissue clearing method, Clear Lipid-exchanged Acrylamide-hybridized Rigid Imaging compatible Tissue hYdrogel (CLARITY), adapted for hMRI validation. Adapting CLARITY to the needs of hMRI is challenging due to poor antibody penetration into large sample volumes and high opacity of aged post mortem human brain tissue. In a pilot experiment we achieved transparency of up to 8 mm-thick and immunohistochemical staining of up to 5 mm-thick post mortem brain tissue by a combination of active and passive clearing, prolonged clearing and staining times. We combined 3D optical imaging of the cleared samples with tailored image processing methods. We demonstrated the feasibility for quantification of neuron density, fiber orientation distribution and cell type classification within a volume with size similar to a typical MRI voxel. The presented combination of MRI, 3D optical microscopy and image processing is a promising tool for validation of MRI-based microstructure estimates.

Single Cell Phenotyping Reveals Heterogeneity Among Hematopoietic Stem Cells Following Infection.

The hematopoietic stem cell (HSC) niche provides essential microenvironmental cues for the production and maintenance of HSCs within the bone marrow. During inflammation, hematopoietic dynamics are perturbed, but it is not known whether changes to the HSC-niche interaction occur as a result. We visualize HSCs directly in vivo, enabling detailed analysis of the 3D niche dynamics and migration patterns in murine bone marrow following Trichinella spiralis infection. Spatial statistical analysis of these HSC trajectories reveals two distinct modes of HSC behavior: (a) a pattern of revisiting previously explored space and (b) a pattern of exploring new space. Whereas HSCs from control donors predominantly follow pattern (a), those from infected mice adopt both strategies. Using detailed computational analyses of cell migration tracks and life-history theory, we show that the increased motility of HSCs following infection can, perhaps counterintuitively, enable mice to cope better in deteriorating HSC-niche microenvironments following infection. Stem Cells 2017;35:2292-2304.

Factor graph analysis of live cell-imaging data reveals mechanisms of cell fate decisions.

MOTIVATION: Cell fate decisions have a strong stochastic component. The identification of the underlying mechanisms therefore requires a rigorous statistical analysis of large ensembles of single cells that were tracked and phenotyped over time. RESULTS: We introduce a probabilistic framework for testing elementary hypotheses on dynamic cell behavior using time-lapse cell-imaging data. Factor graphs, probabilistic graphical models, are used to properly account for cell lineage and cell phenotype information. Our model is applied to time-lapse movies of murine granulocyte-macrophage progenitor (GMP) cells. It decides between competing hypotheses on the mechanisms of their differentiation. Our results theoretically substantiate previous experimental observations that lineage instruction, not selection is the cause for the differentiation of GMP cells into mature monocytes or neutrophil granulocytes. AVAILABILITY AND IMPLEMENTATION: The Matlab source code is available at http://treschgroup.de/Genealogies.html.

In vivo time-lapse imaging shows diverse niche engagement by quiescent and naturally activated hematopoietic stem cells.

Hematopoietic stem cells (HSCs) maintain the turnover of mature blood cells during steady state and in response to systemic perturbations such as infections. Their function critically depends on complex signal exchanges with the bone marrow (BM) microenvironment in which they reside, but the cellular mechanisms involved in HSC-niche interactions and regulating HSC function in vivo remain elusive. We used a natural mouse parasite, Trichinella spiralis, and multipoint intravital time-lapse confocal microscopy of mouse calvarium BM to test whether HSC-niche interactions may change when hematopoiesis is perturbed. We find that steady-state HSCs stably engage confined niches in the BM whereas HSCs harvested during acute infection are motile and therefore interact with larger niches. These changes are accompanied by increased long-term repopulation ability and expression of CD44 and CXCR4. Administration of a CXCR4 antagonist affects the duration of HSC-niche interactions. These findings suggest that HSC-niche interactions may be modulated during infection.

A Standardized Clinical Data Harmonization Pipeline for Scalable AI Application Deployment (FHIR-DHP): Validation and Usability Study.

BACKGROUND: Increasing digitalization in the medical domain gives rise to large amounts of health care data, which has the potential to expand clinical knowledge and transform patient care if leveraged through artificial intelligence (AI). Yet, big data and AI oftentimes cannot unlock their full potential at scale, owing to nonstandardized data formats, lack of technical and semantic data interoperability, and limited cooperation between stakeholders in the health care system. Despite the existence of standardized data formats for the medical domain, such as Fast Healthcare Interoperability Resources (FHIR), their prevalence and usability for AI remain limited. OBJECTIVE: In this paper, we developed a data harmonization pipeline (DHP) for clinical data sets relying on the common FHIR data standard. METHODS: We validated the performance and usability of our FHIR-DHP with data from the Medical Information Mart for Intensive Care IV database. RESULTS: We present the FHIR-DHP workflow in respect of the transformation of "raw" hospital records into a harmonized, AI-friendly data representation. The pipeline consists of the following 5 key preprocessing steps: querying of data from hospital database, FHIR mapping, syntactic validation, transfer of harmonized data into the patient-model database, and export of data in an AI-friendly format for further medical applications. A detailed example of FHIR-DHP execution was presented for clinical diagnoses records. CONCLUSIONS: Our approach enables the scalable and needs-driven data modeling of large and heterogenous clinical data sets. The FHIR-DHP is a pivotal step toward increasing cooperation, interoperability, and quality of patient care in the clinical routine and for medical research.

Simulating Local Deformations in the Human Cortex Due to Blood Flow-Induced Changes in Mechanical Tissue Properties: Impact on Functional Magnetic Resonance Imaging.

Investigating human brain tissue is challenging due to the complexity and the manifold interactions between structures across different scales. Increasing evidence suggests that brain function and microstructural features including biomechanical features are related. More importantly, the relationship between tissue mechanics and its influence on brain imaging results remains poorly understood. As an important example, the study of the brain tissue response to blood flow could have important theoretical and experimental consequences for functional magnetic resonance imaging (fMRI) at high spatial resolutions. Computational simulations, using realistic mechanical models can predict and characterize the brain tissue behavior and give us insights into the consequent potential biases or limitations of in vivo, high-resolution fMRI. In this manuscript, we used a two dimensional biomechanical simulation of an exemplary human gyrus to investigate the relationship between mechanical tissue properties and the respective changes induced by focal blood flow changes. The model is based on the changes in the brain's stiffness and volume due to the vasodilation evoked by neural activity. Modeling an exemplary gyrus from a brain atlas we assessed the influence of different potential mechanisms: (i) a local increase in tissue stiffness (at the level of a single anatomical layer), (ii) an increase in local volume, and (iii) a combination of both effects. Our simulation results showed considerable tissue displacement because of these temporary changes in mechanical properties. We found that the local volume increase causes more deformation and consequently higher displacement of the gyrus. These displacements introduced considerable artifacts in our simulated fMRI measurements. Our results underline the necessity to consider and characterize the tissue displacement which could be responsible for fMRI artifacts.

Automatic Lung Segmentation and Quantification of Aeration in Computed Tomography of the Chest Using 3D Transfer Learning.

BACKGROUND: Identification of lung parenchyma on computer tomographic (CT) scans in the research setting is done semi-automatically and requires cumbersome manual correction. This is especially true in pathological conditions, hindering the clinical application of aeration compartment (AC) analysis. Deep learning based algorithms have lately been shown to be reliable and time-efficient in segmenting pathologic lungs. In this contribution, we thus propose a novel 3D transfer learning based approach to quantify lung volumes, aeration compartments and lung recruitability. METHODS: Two convolutional neural networks developed for biomedical image segmentation (uNet), with different resolutions and fields of view, were implemented using Matlab. Training and evaluation was done on 180 scans of 18 pigs in experimental ARDS (u2Net Pig ) and on a clinical data set of 150 scans from 58 ICU patients with lung conditions varying from healthy, to COPD, to ARDS and COVID-19 (u2Net Human ). One manual segmentations (MS) was available for each scan, being a consensus by two experts. Transfer learning was then applied to train u2Net Pig on the clinical data set generating u2Net Transfer . General segmentation quality was quantified using the Jaccard index (JI) and the Boundary Function score (BF). The slope between JI or BF and relative volume of non-aerated compartment (S JI and S BF , respectively) was calculated over data sets to assess robustness toward non-aerated lung regions. Additionally, the relative volume of ACs and lung volumes (LV) were compared between automatic and MS. RESULTS: On the experimental data set, u2Net Pig resulted in JI = 0.892 [0.88 : 091] (median [inter-quartile range]), BF = 0.995 [0.98 : 1.0] and slopes S JI = -0.2 {95% conf. int. -0.23 : -0.16} and S BF = -0.1 {-0.5 : -0.06}. u2Net Human showed similar performance compared to u2Net Pig in JI, BF but with reduced robustness S JI = -0.29 {-0.36 : -0.22} and S BF = -0.43 {-0.54 : -0.31}. Transfer learning improved overall JI = 0.92 [0.88 : 0.94], P < 0.001, but reduced robustness S JI = -0.46 {-0.52 : -0.40}, and affected neither BF = 0.96 [0.91 : 0.98] nor S BF = -0.48 {-0.59 : -0.36}. u2Net Transfer improved JI compared to u2Net Human in segmenting healthy (P = 0.008), ARDS (P < 0.001) and COPD (P = 0.004) patients but not in COVID-19 patients (P = 0.298). ACs and LV determined using u2Net Transfer segmentations exhibited < 5% volume difference compared to MS. CONCLUSION: Compared to manual segmentations, automatic uNet based 3D lung segmentation provides acceptable quality for both clinical and scientific purposes in the quantification of lung volumes, aeration compartments, and recruitability.

How to predict relapse in leukemia using time series data: A comparative in silico study.

Risk stratification and treatment decisions for leukemia patients are regularly based on clinical markers determined at diagnosis, while measurements on system dynamics are often neglected. However, there is increasing evidence that linking quantitative time-course information to disease outcomes can improve the predictions for patient-specific treatment responses. We designed a synthetic experiment simulating response kinetics of 5,000 patients to compare different computational methods with respect to their ability to accurately predict relapse for chronic and acute myeloid leukemia treatment. Technically, we used clinical reference data to first fit a model and then generate de novo model simulations of individual patients' time courses for which we can systematically tune data quality (i.e. measurement error) and quantity (i.e. number of measurements). Based hereon, we compared the prediction accuracy of three different computational methods, namely mechanistic models, generalized linear models, and deep neural networks that have been fitted to the reference data. Reaching prediction accuracies between 60 and close to 100%, our results indicate that data quality has a higher impact on prediction accuracy than the specific choice of the particular method. We further show that adapted treatment and measurement schemes can considerably improve the prediction accuracy by 10 to 20%. Our proof-of-principle study highlights how computational methods and optimized data acquisition strategies can improve risk assessment and treatment of leukemia patients.

Domain-specific cues improve robustness of deep learning-based segmentation of CT volumes.

Machine learning has considerably improved medical image analysis in the past years. Although data-driven approaches are intrinsically adaptive and thus, generic, they often do not perform the same way on data from different imaging modalities. In particular computed tomography (CT) data poses many challenges to medical image segmentation based on convolutional neural networks (CNNs), mostly due to the broad dynamic range of intensities and the varying number of recorded slices of CT volumes. In this paper, we address these issues with a framework that adds domain-specific data preprocessing and augmentation to state-of-the-art CNN architectures. Our major focus is to stabilise the prediction performance over samples as a mandatory requirement for use in automated and semi-automated workflows in the clinical environment. To validate the architecture-independent effects of our approach we compare a neural architecture based on dilated convolutions for parallel multi-scale processing (a modified Mixed-Scale Dense Network: MS-D Net) to traditional scaling operations (a modified U-Net). Finally, we show that an ensemble model combines the strengths across different individual methods. Our framework is simple to implement into existing deep learning pipelines for CT analysis. It performs well on a range of tasks such as liver and kidney segmentation, without significant differences in prediction performance on strongly differing volume sizes and varying slice thickness. Thus our framework is an essential step towards performing robust segmentation of unknown real-world samples.

Ontology patterns for the representation of quality changes of cells in time.

BACKGROUND: Cell tracking experiments, based on time-lapse microscopy, have become an important tool in biomedical research. The goal is the reconstruction of cell migration patterns, shape and state changes, and, comprehensive genealogical information from these data. This information can be used to develop process models of cellular dynamics. However, so far there has been no structured, standardized way of annotating and storing the tracking results, which is critical for comparative analysis and data integration. The key requirement to be satisfied by an ontology is the representation of a cell's change over time. Unfortunately, popular ontology languages, such as Web Ontology Language (OWL), have limitations for the representation of temporal information. The current paper addresses the fundamental problem of modeling changes of qualities over time in biomedical ontologies specified in OWL. RESULTS: The presented analysis is a result of the lessons learned during the development of an ontology, intended for the annotation of cell tracking experiments. We present, discuss and evaluate various representation patterns for specifying cell changes in time. In particular, we discuss two patterns of temporally changing information: n-ary relation reification and 4d fluents. These representation schemes are formalized within the ontology language OWL and are aimed at the support for annotation of cell tracking experiments. We analyze the performance of each pattern with respect to standard criteria used in software engineering and data modeling, i.e. simplicity, scalability, extensibility and adequacy. We further discuss benefits, drawbacks, and the underlying design choices of each approach. CONCLUSIONS: We demonstrate that patterns perform differently depending on the temporal distribution of modeled information. The optimal model can be constructed by combining two competitive approaches. Thus, we demonstrate that both reification and 4d fluents patterns can work hand in hand in a single ontology. Additionally, we have found that 4d fluents can be reconstructed by two patterns well known in the computer science community, i.e. state modeling and actor-role pattern.

Multi-scale imaging and analysis identify pan-embryo cell dynamics of germlayer formation in zebrafish.

The coordination of cell movements across spatio-temporal scales ensures precise positioning of organs during vertebrate gastrulation. Mechanisms governing such morphogenetic movements have been studied only within a local region, a single germlayer or in whole embryos without cell identity. Scale-bridging imaging and automated analysis of cell dynamics are needed for a deeper understanding of tissue formation during gastrulation. Here, we report pan-embryo analyses of formation and dynamics of all three germlayers simultaneously within a developing zebrafish embryo. We show that a distinct distribution of cells in each germlayer is established during early gastrulation via cell movement characteristics that are predominantly determined by their position in the embryo. The differences in initial germlayer distributions are subsequently amplified by a global movement, which organizes the organ precursors along the embryonic body axis, giving rise to the blueprint of organ formation. The tools and data are available as a resource for the community.

Cell-accurate optical mapping across the entire developing heart.

Organogenesis depends on orchestrated interactions between individual cells and morphogenetically relevant cues at the tissue level. This is true for the heart, whose function critically relies on well-ordered communication between neighboring cells, which is established and fine-tuned during embryonic development. For an integrated understanding of the development of structure and function, we need to move from isolated snap-shot observations of either microscopic or macroscopic parameters to simultaneous and, ideally continuous, cell-to-organ scale imaging. We introduce cell-accurate three-dimensional Ca2+-mapping of all cells in the entire electro-mechanically uncoupled heart during the looping stage of live embryonic zebrafish, using high-speed light sheet microscopy and tailored image processing and analysis. We show how myocardial region-specific heterogeneity in cell function emerges during early development and how structural patterning goes hand-in-hand with functional maturation of the entire heart. Our method opens the way to systematic, scale-bridging, in vivo studies of vertebrate organogenesis by cell-accurate structure-function mapping across entire organs.

P2Y(1) receptor mediated neuronal fibre outgrowth in organotypic brain slice co-cultures.

Extracellular purines have multiple functional roles in development, plastic remodelling, and regeneration of the CNS by stimulating certain P2X/Y receptor (R) subtypes. In the present study we elucidated the involvement of P2YRs in neuronal fibre outgrowth in the developing nervous system. We particularly focused on the P2Y1R subtype and the dopaminergic system, respectively. For this purpose, we used organotypic slice co-cultures consisting of the ventral tegmental area/substantia nigra (VTA/SN) and the prefrontal cortex (PFC). After detecting the presence of the P2Y1R in VTA/SN, PFC, and on outgrowing fibres in the border region (e.g. on glial processes) connecting both brain slices, we could show that pharmacological modulation of the receptor influenced neuronal fibre outgrowth. Biocytin-tracing and tyrosine hydroxylase-immunolabelling together with quantitative image analysis revealed a significant increase in fibre growth in the border region of the co-cultures after treatment with ADPßS (P2Y1,12,13R agonist). The observed stimulatory potential of ADPßS was inhibited by pre-treatment with the P2X/YR antagonist PPADS. In P2Y1R knockout (P2Y1R(-/-)) mice, the ADPßS-induced stimulatory effect was absent, while growth was significantly enhanced in the co-cultures of the respective wild-type. This observation was confirmed in entorhino-hippocampal co-cultures, an example of a different projection system, expressing the P2Y1R. Using wortmannin and PD98059 we further showed that PI3K/Akt and MAPK/ERK cascades are involved in the mechanism underlying ADPßS-induced fibre growth. In conclusion, the data of this study clearly indicate that activation of the P2Y1R stimulates fibre growth and thereby emphasises the general role of this particular receptor subtype during development and regeneration.

Nimodipine enhances neurite outgrowth in dopaminergic brain slice co-cultures.

Calcium ions (Ca(2+)) play important roles in neuroplasticity and the regeneration of nerves. Intracellular Ca(2+) concentrations are regulated by Ca(2+) channels, among them L-type voltage-gated Ca(2+) channels, which are inhibited by dihydropyridines like nimodipine. The purpose of this study was to investigate the effect of nimodipine on neurite growth during development and regeneration. As an appropriate model to study neurite growth, we chose organotypic brain slice co-cultures of the mesocortical dopaminergic projection system, consisting of the ventral tegmental area/substantia nigra and the prefrontal cortex from neonatal rat brains. Quantification of the density of the newly built neurites in the border region (region between the two cultivated slices) of the co-cultures revealed a growth promoting effect of nimodipine at concentrations of 0.1µM and 1µM that was even more pronounced than the effect of the growth factor NGF. This beneficial effect was absent when 10µM nimodipine were applied. Toxicological tests revealed that the application of nimodipine at this higher concentration slightly induced caspase 3 activation in the cortical part of the co-cultures, but did neither affect the amount of lactate dehydrogenase release or propidium iodide uptake nor the ratio of bax/bcl-2. Furthermore, the expression levels of different genes were quantified after nimodipine treatment. The expression of Ca(2+) binding proteins, immediate early genes, glial fibrillary acidic protein, and myelin components did not change significantly after treatment, indicating that the regulation of their expression is not primarily involved in the observed nimodipine mediated neurite growth. In summary, this study revealed for the first time a neurite growth promoting effect of nimodipine in the mesocortical dopaminergic projection system that is highly dependent on the applied concentrations.

Mesenchymal stem cells support neuronal fiber growth in an organotypic brain slice co-culture model.

Mesenchymal stem cells (MSCs) have been identified as promising candidates for neuroregenerative cell therapies. However, the impact of different isolation procedures on the functional and regenerative characteristics of MSC populations has not been studied thoroughly. To quantify these differences, we directly compared classically isolated bulk bone marrow-derived MSCs (bulk BM-MSCs) to the subpopulation Sca-1(+)Lin(-)CD45(-)-derived MSCs(-) (SL45-MSCs), isolated by fluorescence-activated cell sorting from bulk BM-cell suspensions. Both populations were analyzed with respect to functional readouts, that are, frequency of fibroblast colony forming units (CFU-f), general morphology, and expression of stem cell markers. The SL45-MSC population is characterized by greater morphological homogeneity, higher CFU-f frequency, and significantly increased nestin expression compared with bulk BM-MSCs. We further quantified the potential of both cell populations to enhance neuronal fiber growth, using an ex vivo model of organotypic brain slice co-cultures of the mesocortical dopaminergic projection system. The MSC populations were cultivated underneath the slice co-cultures without direct contact using a transwell system. After cultivation, the fiber density in the border region between the two brain slices was quantified. While both populations significantly enhanced fiber outgrowth as compared with controls, purified SL45-MSCs stimulated fiber growth to a larger degree. Subsequently, we analyzed the expression of different growth factors in both cell populations. The results show a significantly higher expression of brain-derived neurotrophic factor (BDNF) and basic fibroblast growth factor in the SL45-MSCs population. Altogether, we conclude that MSC preparations enriched for primary MSCs promote neuronal regeneration and axonal regrowth, more effectively than bulk BM-MSCs, an effect that may be mediated by a higher BDNF secretion.

Elucidating functional heterogeneity in hematopoietic progenitor cells: a combined experimental and modeling approach.

A detailed understanding of the mechanisms maintaining the hierarchical balance of cell types in hematopoiesis will be important for the therapeutic manipulation of normal and leukemic cells. Mathematical modeling is expected to make an important contribution to this area, but the iterative development of increasingly accurate models will rely on repeated validation using experimental data of sufficient resolution to distinguish between alternative model scenarios. The multipotent hematopoietic progenitor FDCP-Mix cells maintain a hierarchy from self-renewal to post-mitotic differentiation in vitro and are accessible to detailed analysis. Here, we report the development of a combined mathematical modeling and experimental approach to study the principles underlying heterogeneity in FDCP-Mix cultures. We adapt a single-cell based model of hematopoiesis to the conditions of cell culture and describe an association between proliferative history and phenotype of FDCP-Mix cells. While data derived from population studies are incapable of distinguishing between three mechanistically different model scenarios, statistical analysis of single cell tracking data provides a resolution sufficient to select one of them. This scenario favors differences between granulocytic and monocytic lineage with respect to their proliferative behavior and death rates as a mechanistic explanation for the observed heterogeneity. Our results demonstrate the power of a combined experimental/modeling approach in which single cell fate analysis is the key to revealing regulatory principles at the cellular level.

High-speed panoramic light-sheet microscopy reveals global endodermal cell dynamics.

The ever-increasing speed and resolution of modern microscopes make the storage and post-processing of images challenging and prevent thorough statistical analyses in developmental biology. Here, instead of deploying massive storage and computing power, we exploit the spherical geometry of zebrafish embryos by computing a radial maximum intensity projection in real time with a 240-fold reduction in data rate. In our four-lens selective plane illumination microscope (SPIM) setup the development of multiple embryos is recorded in parallel and a map of all labelled cells is obtained for each embryo in <10 s. In these panoramic projections, cell segmentation and flow analysis reveal characteristic migration patterns and global tissue remodelling in the early endoderm. Merging data from many samples uncover stereotypic patterns that are fundamental to endoderm development in every embryo. We demonstrate that processing and compressing raw image data in real time is not only efficient but indispensable for image-based systems biology.