Figure 3 - uploaded by Jochen Gehrig
Content may be subject to copyright.
Automated dorsal imaging of zebrafish embryonic brains. (A) Laser scanning image of milled keel-shaped cavities for facilitating orientation of zebrafish embryos (depth, 800 µm; side wall inclination, 60°; and length, 5 mm) and serving as template to generate the silicone tool shown in panel B. (B) Photograph of polydimethylsiloxane replica of the array of cavities used to generate an array of 96 agarose molds. (C) Schematic depiction of a single well with ventrally oriented embryo within an agarose mold. The bottomless well plate, agarose molds containing a ventrally oriented embryo, and the imaging direction are indicated. Mold dimensions and agarose thickness are as in panel A. Drawing is not to scale. (D–F) Dorsal views of a single embryo lying within the agarose mold. (D) Prescreen data acquired with 2.5× objective. Rectangle and asterisk are as in Figure 1C. (E) Extended-focus image of 120 z-slices (4-µm slice distance) of the same embryo, bright-field view. (F) Extended-focus images of 120 z-slices (4-µm slice distance) in the GFP channel previously deconvolved using Huygens Professional software (see also Supplementary Movie 1). Scale bars: 320 µm (D) and 80 µm (E and F).
Source publication
The development of automated microscopy platforms has enabled large-scale observation of biological processes, thereby complementing genome scale biochemical techniques. However, commercially available systems are restricted either by fixed-field-of-views, leading to potential omission of features of interest, or by low-resolution data of whole obj...
Contexts in source publication
Context 1
... to the thickness and bilateral body plan of the zebrafish embryo, visualization of many organs is hindered by lateral orientation. Therefore, a simple keel-shaped mold geometry was designed that allows tilt-free ventral orien- tation of 2-3 day-old embryos, facilitating the imaging of dorsal views when using an inverted microscope (Figure 3, A-C). An array of 96 molds was milled into a polymeth- ylmethacrylate dish and replicated by casting a silicone elastomer (Figure 3, A and B). ...
Context 2
... a simple keel-shaped mold geometry was designed that allows tilt-free ventral orien- tation of 2-3 day-old embryos, facilitating the imaging of dorsal views when using an inverted microscope (Figure 3, A-C). An array of 96 molds was milled into a polymeth- ylmethacrylate dish and replicated by casting a silicone elastomer (Figure 3, A and B). This silicone replica was used again as a template to generate an array of 96 molds into a thin layer of agarose poured into the lid of a 96-well plate. ...
Context 3
... silicone replica was used again as a template to generate an array of 96 molds into a thin layer of agarose poured into the lid of a 96-well plate. To minimize water movements and thus stabilize embryos, a bottomless 96-well plate was pressed onto the array of molds ( Figure 3C). The design of the agarose plate prevents its usage in drug or toxicological screens, as liquid can diffuse through the agarose. ...
Context 4
... a different template was implemented in the algorithm. This modified algorithm detected the position of GFP-positive embryonic brains with an accuracy of 98.3% (n = 366) ( Figure 3D). In a model experiment, 48 hpf embryos of the ETvmat2:gfp transgenic line were ventrally arrayed and prescreen data of dorsal views were acquired. ...
Context 5
... a model experiment, 48 hpf embryos of the ETvmat2:gfp transgenic line were ventrally arrayed and prescreen data of dorsal views were acquired. Subsequently, for GFP-positive embryos, 120 z-slices with a distance of 4 µm were acquired using a 10× objective, giving rise to data sets allowing a much more detailed visualization of the embryonic brain compared with lateral views (Figure 3, E-F; see also Supplementary Figure S2). ...
Context 6
... this end, GFP z-stacks were batch-decon- volved with Huygens Professional software using a theoretical point spread function. This results in high-resolution 3-D data sets visual- izing the spatial organization of GFP-positive cell clusters within the embryonic brain ( Figure 3F and Supplementary Movie S1). Thus, the protocol enables automatic, high- throughput acquisition of high-resolution z-volumes and, in combination with deconvo- lution, gives rise to cellular-resolution 3-D data sets with sufficient detail to analyze the spatial distribution of fluorescence signal within the complexity of a living organism. ...
Context 7
... of dorsal views of embryonic brains were acquired and restored by deconvolution. The partial overlap of CFP and DsRed-T4 expression within the distinct cells could be detected by quantitative colocalization analysis (26) (Supplementary Figure S3). Therefore, the developed toolset in combi- nation with deconvolution allowed the acqui- sition of data of sufficient detail to analyze complex spatial reporter gene expression patterns within the living embryo. ...
Similar publications
Protocol Summary
Here we demonstrate a toolset for automated intelligent high-content screening of whole zebrafish embryos at cellular resolution on a standard wide-field screening microscope. Using custom-developed algorithms, predefined regions of interest, such as the brain, are automatically detected. The regions of interest are subsequently im...
Citations
... A potential shortcoming in the field of comparative phenomics, and across evolutionary developmental biology more generally is an absence of phenotyping approaches that are readily transferable to non-model species of interest (Tills et al., 2018). One of the cornerstones of comparative developmental physiology is the selection of species that are best suited to answer a particular biological question (Krogh, 1929), yet approaches to high-dimensional organismal phenotyping remain constrained to model animals of interest, particularly the zebrafish Danio rerio (Pelkowski et al., 2011;Kalueff et al., 2016;Peravali et al., 2018), nematode worm Caenorhabditis elegans (Yemini et al., 2013;Cornaglia et al., 2015;Olmedo et al., 2015), and the fruit fly Drosophila melanogaster (Chung et al., 2010;Levario et al., 2016). Here, EPTs were effective at characterising high-dimensional functional change in embryos of three species of freshwater gastropod, despite significant differences in their patterns of development. ...
Understanding the links between development and evolution is one of the major challenges of biology. 'Heterochronies', evolutionary alterations in the timings of development are posited as a key mechanism of evolutionary change, but their quantification requires gross simplification of organismal development. Consequently, how changes in event timings influence development more broadly is poorly understood. Here, we measure organismal development as spectra of energy in pixel values of video, creating high-dimensional landscapes integrating development of all visible form and function. This approach we termed 'Energy proxy traits' (EPTs) is applied alongside previously identified heterochronies in three freshwater pulmonate molluscs (Lymnaea stagnalis, Radix balthica and Physella acuta). EPTs were calculated from time-lapse video of embryonic development to construct a continuous functional time series. High-dimensional transitions in phenotype aligned with major sequence heterochronies between species. Furthermore, differences in event timings between conspecifics were associated with changes in high-dimensional phenotypic space. We reveal EPTs as a powerful approach to considering the evolutionary importance of alterations to developmental event timings. Reimagining the phenotype as energy spectra enabled continuous quantification of developmental changes in high-dimensional phenotypic space, rather than measurement of timings of discrete events. This approach has the possibility to transform how we study heterochrony and development more generally.
... This includes advancements in phenotyping technologies for Frontiers in Physiology frontiersin.org 02 common model species including Arabidopsis (Furbank and Tester, 2011;Vanhaeren et al., 2015), the zebrafish Danio rerio (Xu et al., 2010;Pelkowski et al., 2011;Peravali et al., 2018;Spomer et al., 2012), the nematode Caenorhabditis elegans (White et al., 2010;Olmedo et al., 2015) and the fruit fly Drosophila melanogaster (Dagani et al., 2007;Chung et al., 2010;Levario et al., 2016). In a practical sense, phenomics typically takes the form of sensors, combined with some degree of automation, whether analytical such as computer vision pipelines (Tills et al., 2018;Choudhury et al., 2019), or physical such as robotics for processing samples (Yang et al., 2020). ...
The dynamic nature of developing organisms and how they function presents both opportunity and challenge to researchers, with significant advances in understanding possible by adopting innovative approaches to their empirical study. The information content of the phenotype during organismal development is arguably greater than at any other life stage, incorporating change at a broad range of temporal, spatial and functional scales and is of broad relevance to a plethora of research questions. Yet, effectively measuring organismal development, and the ontogeny of physiological regulations and functions, and their responses to the environment, remains a significant challenge. "Phenomics", a global approach to the acquisition of phenotypic data at the scale of the whole organism, is uniquely suited as an approach. In this perspective, we explore the synergies between phenomics and Comparative Developmental Physiology (CDP), a discipline of increasing relevance to understanding sensitivity to drivers of global change. We then identify how organismal development itself provides an excellent model for pushing the boundaries of phenomics, given its inherent complexity, comparably smaller size, relative to adult stages, and the applicability of embryonic development to a broad suite of research questions using a diversity of species. Collection, analysis and interpretation of whole organismal phenotypic data are the largest obstacle to capitalising on phenomics for advancing our understanding of biological systems. We suggest that phenomics within the context of developing organismal form and function could provide an effective scaffold for addressing grand challenges in CDP and phenomics.
... A more do-it-yourself approach involves creation of a stamp to create a trough within agarose to hold the zebrafish embryos in desired dorsal or on-side orientations. The stamps can be fabricated from polydimethylsiloxane (PDMS) [35] , brass [36] , or using 3D printing [37] . In each of these cases, the stamp creates a reproducible array of troughs in agarose, often within a 96-well plate. ...
Fundamental life science and pharmaceutical research are continually striving to provide physiologically relevant context for their biological studies. Zebrafish present an opportunity for high-content screening (HCS) to bring a true in vivo model system to screening studies. Zebrafish embryos and young larvae are an economical, human-relevant model organism that are amenable to both genetic engineering and modification, and direct inspection via microscopy. The use of these organisms entails unique challenges that new technologies are overcoming, including artificial intelligence (AI).
In this perspective article, we describe the state-of-the-art in terms of automated sample handling, imaging, and data analysis with zebrafish during early developmental stages. We highlight advances in orienting the embryos, including the use of robots, microfluidics, and creative multi-well plate solutions. Analyzing the micrographs in a fast, reliable fashion that maintains the anatomical context of the fluorescently labeled cells is a crucial step. Existing software solutions range from AI-driven commercial solutions to bespoke analysis algorithms. Deep learning appears to be a critical tool that researchers are only beginning to apply, but already facilitates many automated steps in the experimental workflow. Currently, such work has permitted the cellular quantification of multiple cell types in vivo, including stem cell responses to stress and drugs, neuronal myelination and macrophage behavior during inflammation and infection. We evaluate pro and cons of proprietary versus open-source methodologies for combining technologies into fully automated workflows of zebrafish studies. Zebrafish are poised to charge into HCS with ever-greater presence, bringing a new level of physiological context.
... HCS approaches couple automated multiparametric microscope-based image acquisition with powerful image processing and analysis algorithms for high throughput automated feature detection (Peravali et al. 2011), quantitation and classification. In this section, we focus on describing the data acquisition instrumentation (hardware) necessary for conducting HCS assays with the zebrafish model. ...
The rise of nanotechnology has led to concerns about its potential risks to the environment and human health. Nanotoxicology, the study on the toxicological effects of nanoparticles, aims to address these concerns through the use of in vitro and in vivo toxicity tests. Although traditional toxicity testing relies primarily on in vivo models such as rodents, there is a shift towards the use of in vitro and alternative models in recent years. While useful for determining tissue-specific toxicity, in vitro cell lines fail to provide information at the whole organismal level. Zebrafish have emerged in the past two decades as a powerful alternative in vivo toxicity model. Their small size, transparency during embryonic and larval stages, rapid development and availability of transgenic zebrafish reporter lines make them highly suitable for high throughput screening (HTS) assays. Also, recent advances in automation, robotic handling and artificial intelligence (AI) assisted data analysis have further enhanced the screening capability. With the establishment of property-toxicity relationships, safer-by-design of nanomaterials have become feasible. This chapter summarizes the key features of the zebrafish toxicity model and HTS which contribute towards its suitability as a nanotoxicity model and highlights applications of zebrafish in nanotoxicity testing.
... Additionally, image analysis is bound to proprietary modules (Nikon, 2020;Zeiss, 2020), which compromises reproducibility and limits flexibility. Open source solutions are available for Intelligent Microscopy but they either consist in charting a 2D (Carro et al., 2015) or 3D (Peravali et al., 2011;Pinkard et al., 2016) sample so as to restrict the imaging area, or they are designed for target detection from still images (Tischer et al., 2014;Booth et al., 2018;Micro-Manager, 2020) (even though live microscopy can be performed at the detected positions). An exception is Micro-pilot (Conrad et al., 2011), a pioneering project designed for the real-time detection of cell mitosis (and their subsequent FRAP manipulation), but this software is difficult to adapt to other applications since it was crafted for this specific application and developed in a low level language for a set of very specific microscopes. ...
We developed AutoscanJ, a suite of ImageJ scripts enabling to image targets of interest by automatically driving a motorized microscope at the corresponding locations. For live samples, our software can sequentially detect biological events from their onset and further image them at high resolution, an action that would be impractical by user operation. For fixed samples, the software can dramatically reduce the amount of data acquired and the acquisition duration in situations where statistically few targets of interest are observed per field of view. AutoScanJ is compatible with motorized fluorescence microscopes controlled by Leica LAS AF/X or Micro-Manager. The software is straightforward to set up and new custom image analysis workflows to detect targets of interest can be simply implemented and shared with minimal efforts as independent ImageJ macro functions. We illustrate five different application scenarios with the system ranging from samples fixed on micropatterned surfaces to live cells undergoing several rounds of division. The target detection functions for these applications are provided and can be used as a starting point and a source of inspiration for new applications. Overall, AutoScanJ helps to optimize microscope usage by autonomous operation, and it opens up new experimental avenues by enabling the real-time detection and selective imaging of transient events in live microscopy.
... Various digital image analysis tools have been proposed to partially or fully automate cardiac screening in fruit fly and zebrafish [17,19,20,34,[39][40][41][42][43][44][45][46][47]. Berh et al. [19] introduced an automated in-vivo heartbeat detection algorithm for D. melanogaster pupae based on frustrated total internal reflection (FTIR). ...
In this paper, the heartbeat parameters of small model organisms, i.e. Drosophila melanogaster (fruit fly) and Danio rerio (zebrafish), were quantified in-vivo in intact larvae using microfluidics and a novel MATLAB-based software. Among different developmental stages of flies and zebrafish, the larval stage is privileged due to biological maturity, optical accessibility, and the myogenic nature of the heart. Conventional methods for parametric quantification of heart activities are complex and mostly done on dissected, irreversibly immobilized, or anesthetized larvae. Microfluidics has helped with reversible immobilization without the need for anesthesia, but heart monitoring is still done manually due to challenges associated with the movement of floating organs and cardiac interruptions. In our MATLAB software applied to videos recorded in microfluidic-based whole-organism assays, we have used image segmentation to automatically detect the heart and extract the heartbeat signal based on pixel intensity variations of the most contractile region of the heart tube. The smoothness priors approach (SPA) was applied to remove the undesired low-frequency noises caused by environmental light changes or heart movement. Heart rate and arrhythmicity were automatically measured from the detrended heartbeat signal while other parameters including end-diastolic and end-systolic diameters, shortening distance, shortening time, fractional shortening, and shortening velocity were quantified for the first time in intact larvae, using M-mode images under bright field microscopy. The software was able to detect more than 94% of the heartbeats and the cardiac arrests in intact Drosophila larvae. Our user-friendly software enables in-vivo quantification of D. melanogaster and D. rerio larval heart functions in microfluidic devices, with the potential to be applied to other biological models and used for automatic screening of drugs and alleles that affect their heart.
... Wild-type (AB/WIK strain) adult zebrafish ( n = 39) were obtained from an adult zebrafish colony housed at the University of Washington. The AB/WIK strain has commonly been used in developmental and physiological zebrafish studies (e.g., Fisher et al., 2003, Bahary et al., 2004, Shima et al., 2009, Peravali et al., 2011, Kimelman, 2016, Staudacher et al., 2019. Animals used in experiments were from three different mean age groups based on the time difference between date of hatching and date of testing that included 13-month, 20-month, and 37-month-old mean-age groups. ...
Age-related hearing loss (ARHL), also known as presbycusis, is a widespread and debilitating condition impacting many older adults. Conventionally, researchers utilize mammalian model systems or human cadaveric tissue to study ARHL pathology. Recently, the zebrafish has become an effective and tractable model system for a wide variety of genetic and environmental auditory insults, but little is known about the incidence or extent of ARHL in zebrafish and other non-mammalian models. Here, we evaluated whether zebrafish exhibit age-related loss in auditory sensitivity. The auditory sensitivity of adult wild-type zebrafish (AB/WIK strain) from three adult age subgroups (13-month, 20-month, and 37-month) was characterized using the auditory evoked potential (AEP) recording technique. AEPs were elicited using pure tone stimuli (115-4500 Hz) presented via an underwater loudspeaker and recorded using shielded subdermal metal electrodes. Based on measures of sound pressure and particle acceleration, the mean AEP thresholds of 37-month-old fish [mean sound pressure level (SPL) = 122.2 dB ± 2.2 dB SE re: 1 μPa; mean particle acceleration level (PAL) = -27.5 ± 2.3 dB SE re: 1 ms⁻²] were approximately 9 dB higher than that of 20-month-old fish [(mean SPL = 113.1 ± 2.7 dB SE re: 1 μPa; mean PAL = -37.2 ± 2.8 dB re: 1 ms⁻²; p = 0.007)] and 6 dB higher than that of 13-month-old fish [(mean SPL = 116.3 ± 2.5 dB SE re: 1 μPa; mean PAL = -34.1 ± 2.6 dB SE re: 1 ms⁻²; p = 0.052)]. Lowest AEP thresholds for all three age groups were generally between 800 Hz and 1850 Hz, with no evidence for frequency-specific age-related loss. Our results suggest that zebrafish undergo age-related loss in auditory sensitivity, but the form and magnitude of loss is markedly different than in mammals, including humans. Future work is needed to further describe the incidence and extent of ARHL across vertebrate groups and to determine which, if any, ARHL mechanisms may be conserved across vertebrates to support meaningful comparative/translational studies.
... At the scale of a microscope, small organisms are large three-dimensional objects, and thus more complicated to image than flat cell cultures. In particular, without adapted mounting strategies, these specimens will adopt random positions and orientations in wells of microtiter plates, which complicates the standardized imaging of tissues or organs of interests at high-throughput (Burns et al. 2005;Gehrig et al. 2009;Peravali et al. 2011;Spomer et al. 2012). Dedicated sample mounting strategies compatible with high-throughput imaging are thus necessary to assure the reliable orientation and positioning of specimens (Alessandri et al. 2017;Chang et al. 2012;Fuad et al. 2018;Pardo-Martin et al. 2010;Popova et al. 2018;Wittbrodt et al. 2014). ...
... Finally depending on the outcome of the analysis (the feedback), a new sequence of events can be executed (displacement of the objective, high-resolution acquisition, adjustment of the illumination). This procedure can then be repeated for multiple samples e.g. for the automated imaging of a full well plate (Pandey et al. 2019;Peravali et al. 2011;Spomer et al. 2012), or iteratively for the same specimen to adapt the acquisition over time (Wang et al. 2014). A widespread example of feedback microscopy is the autofocus, traditionally available on most commercial microscopes. ...
... Besides cell cultures, organoids (Lukonin et al. 2020) and small organisms such as zebrafish larvae have also been imaged at high-resolution with high-throughput, thanks to feedback microscopy with such a prescan/rescan strategy. (Peravali et al. 2011) thus demonstrated the imaging of the zebrafish head region detected by template matching, (Pandey et al. 2019) described the high-resolution acquisition of the fluorescently-labelled kidney in the Tg(wt1b:EGFP) transgenic line, localized by centre of mass, while (Spomer et al. 2012) used intensity-based thresholding to identify and image the heart-region. ...
This dissertation reports the design, development and benchmarking of novel research software inspired by the field of computer vision and data-science. The aim was to create versatile and robust solutions tailored to the requirements of microscopy-based phenotypic screening studies in small model organisms. The resulting software tools address various steps of the screening workflow, from manual ground-truth annotations, automated detection of regions of interest, targeted imaging of specific tissues or organs using feedback microscopy, image-classification, and interactive data exploration. Importantly, the tools are generic by design and were benchmarked on several microscopy datasets of zebrafish larvae and medaka embryos. They are particularly suitable for phenotypic screening studies at the tissue- and organ specific level. The software is easy-to-use and readily accessible to biomedical researchers with little to no prior knowledge of computer vision or image-processing. The tools are integrated in common scientific image-analysis packages and accompanied by extensive documentation in the form of articles in academic journals, readme files accompanying the source codes and online video tutorials. To foster their distribution and the inspiration of derived work, most of the underlying source code is available online in open-source repositories.
... Previous implementations of template matching [14,23,24] do not allow to use different templates for object detections, neither do they provide a way to prevent multiple detections of the same object, which motivated our implementation of a NMS strategy. A comparison to previous template matching tools available in common end-user bioimage analysis software is given in Table 1. ...
... To improve computing efficiency, the parallelization of template searches or GPU computing with OpenCV could be explored. Finally, the demonstrated template matching tools could also facilitate feedback microscopy applications by interfacing it with the control software of automated microscopes, thus enabling the automated acquisition of ROI for tracking or automated zooming-in on target structures without manual intervention [21,23]. Additional file 4: Figure S1. ...
Background
The localization of objects of interest is a key initial step in most image analysis workflows. For biomedical image data, classical image-segmentation methods like thresholding or edge detection are typically used. While those methods perform well for labelled objects, they are reaching a limit when samples are poorly contrasted with the background, or when only parts of larger structures should be detected. Furthermore, the development of such pipelines requires substantial engineering of analysis workflows and often results in case-specific solutions. Therefore, we propose a new straightforward and generic approach for object-localization by template matching that utilizes multiple template images to improve the detection capacity.
Results
We provide a new implementation of template matching that offers higher detection capacity than single template approach, by enabling the detection of multiple template images. To provide an easy-to-use method for the automatic localization of objects of interest in microscopy images, we implemented multi-template matching as a Fiji plugin, a KNIME workflow and a python package. We demonstrate its application for the localization of entire, partial and multiple biological objects in zebrafish and medaka high-content screening datasets. The Fiji plugin can be installed by activating the Multi-Template-Matching and IJ-OpenCV update sites. The KNIME workflow is available on nodepit and KNIME Hub. Source codes and documentations are available on GitHub (https://github.com/multi-template-matching).
Conclusion
The novel multi-template matching is a simple yet powerful object-localization algorithm, that requires no data-pre-processing or annotation. Our implementation can be used out-of-the-box by non-expert users for any type of 2D-image. It is compatible with a large variety of applications including, for instance, analysis of large-scale datasets originating from automated microscopy, detection and tracking of objects in time-lapse assays, or as a general image-analysis step in any custom processing pipelines. Using different templates corresponding to distinct object categories, the tool can also be used for classification of the detected regions.
... This involves the use of transgenic lines of fish embryos that express fluorescent protein in a specific cells population of a specific organ. This process facilitates observations of the structure of interest and makes it easier to identify chemicals that modulate gene expression [73]. Some examples are the quantification of neural patterns in the spinal cord of zebrafish, or the detection of chemicals that cause yolk malabsorption [74,75]. ...
Numerous chemicals are used as ingredients by the cosmetics industry and are included in cosmetics formula. Aside from the assessment of their efficacy, the cosmetics industry especially needs to assess the safety of their chemicals for human. Toxicological screening of chemicals is performed with the aim of revealing the potential toxic effect of the tested chemical. Among the potential effects we want to detect, the developmental toxicity of the chemical (teratogenicity), meaning its capability of provoking abnormalities during the embryonic development, is crucial. With respect to the international regulations that forbid the use of animal testing for the safety assessment of cosmetics, the toxicological assessment of chemicals must base on an ensemble of in silico assays, in vitro assays and alternative models based assays. For now, a few alternative methods have been validated in the field of developmental toxicology. The development of new alternative methods is thus required. In addition to the safety assessment, the environmental toxicity assessment is also required. The use of most of cosmetics and personal care products leads to their rejection in waterways after washing and rince. This results in the exposition of some aquatic environments (surface waters and coastal marine environments) to chemicals included in cosmetics and personal care products. Thus, the environmental assessment of cosmetics and of their ingredients requires the knowledge of their toxicity on organisms that are representative of aquatic food chains. In this context, the fish embryo model, which is ethically acceptable according to international regulations, presents a dual advantage for the cosmetics industry. Firstly, as a model representative of aquatic organisms, it is accurate for the environmental assessment of chemicals. Secondly, this model is promising for the assessment of the teratogenic effect of chemicals on human. For this reason, a teratogenicity assessment test is developed. This test is based on the analysis of medaka fish embryos (Oryzias Latipes) at 9 days post fertilization, after balneation in a predetermined concentration of the chemical under study. The analysis of functional and morphological parameters allows to calculate a teratogenicity index, that depends on both rates of dead and malformed embryos. This index allows to to draw a conclusion concerning the teratogenic effect of the chemical.The objective of this project is to automate the teratogenicity test, by automated image and video classification. A first method is developed that aims to automatically detect embryo heart beats from acquired video sequences. This method will allow to calculate the proportion of dead embryos. We then focus on the detection of two common malformations: axial malformations and absence of a swim bladder, based on a machine learning classification. This analysis must be completed by the detection of other malformations so that we can measure the rate of malformed embryos and thus, calculate the teratogenicity index of the tested chemical
