Automatic Segmentation and Cardiac Mechanics Analysis of Evolving Zebrafish Using Deep Learning

Bringing Artificial Intelligence (AI) into Environmental Toxicology Studies: A Perspective of AI-Enabled Zebrafish High-Throughput Screening

The booming development of artificial intelligence (AI) has brought excitement to many research fields that could benefit from its big data analysis capability for causative relationship establishment and knowledge generation. In toxicology studies using zebrafish, the microscopic images and videos that illustrate the developmental stages, phenotypic morphologies, and animal behaviors possess great potential to facilitate rapid hazard assessment and dissection of the toxicity mechanism of environmental pollutants. However, the traditional manual observation approach is both labor-intensive and time-consuming. In this Perspective, we aim to summarize the current AI-enabled image and video analysis tools to realize the full potential of AI. For image analysis, AI-based tools allow fast and objective determination of morphological features and extraction of quantitative information from images of various sorts. The advantages of providing accurate and reproducible results while avoiding human intervention play a critical role in speeding up the screening process. For video analysis, AI-based tools enable the tracking of dynamic changes in both microscopic cellular events and macroscopic animal behaviors. The subtle changes revealed by video analysis could serve as sensitive indicators of adverse outcomes. With AI-based toxicity analysis in its infancy, exciting developments and applications are expected to appear in the years to come.

4D Light-sheet Imaging of Zebrafish Cardiac Contraction

Zebrafish is an intriguing model organism known for its remarkable cardiac regeneration capacity. Studying the contracting heart in vivo is essential for gaining insights into structural and functional changes in response to injuries. However, obtaining high-resolution and high-speed 4-dimensional (4D, 3D spatial + 1D temporal) images of the zebrafish heart to assess cardiac architecture and contractility remains challenging. In this context, an in-house light-sheet microscope (LSM) and customized computational analysis are used to overcome these technical limitations. This strategy, involving LSM system construction, retrospective synchronization, single cell tracking, and user-directed analysis, enables one to investigate the micro-structure and contractile function across the entire heart at the single-cell resolution in the transgenic Tg(myl7:nucGFP) zebrafish larvae. Additionally, we are able to further incorporate microinjection of small molecule compounds to induce cardiac injury in a precise and controlled manner. Overall, this framework allows one to track physiological and pathophysiological changes, as well as the regional mechanics at the single-cell level during cardiac morphogenesis and regeneration.

Identification of side effects of COVID-19 drug candidates on embryogenesis using an integrated zebrafish screening platform

Drug repurposing is an important strategy in COVID-19 treatment, but many clinically approved compounds have not been extensively studied in the context of embryogenesis, thus limiting their administration during pregnancy. Here we used the zebrafish embryo model organism to test the effects of 162 marketed drugs on cardiovascular development. Among the compounds used in the clinic for COVD-19 treatment, we found that Remdesivir led to reduced body size and heart functionality at clinically relevant doses. Ritonavir and Baricitinib showed reduced heart functionality and Molnupiravir and Baricitinib showed effects on embryo activity. Sabizabulin was highly toxic at concentrations only 5 times higher than Cmax and led to a mean mortality of 20% at Cmax. Furthermore, we tested if zebrafish could be used as a model to study inflammatory response in response to spike protein treatment and found that Remdesivir, Ritonavir, Molnupiravir, Baricitinib as well as Sabizabulin counteracted the inflammatory response related gene expression upon SARS-CoV-2 spike protein treatment. Our results show that the zebrafish allows to study immune-modulating properties of COVID-19 compounds and highlights the need to rule out secondary defects of compound treatment on embryogenesis. All results are available on a user friendly web-interface https://share.streamlit.io/alernst/covasc_dataapp/main/CoVasc_DataApp.py that provides a comprehensive overview of all observed phenotypic effects and allows personalized search on specific compounds or group of compounds. Furthermore, the presented platform can be expanded for rapid detection of developmental side effects of new compounds for treatment of COVID-19 and further viral infectious diseases.

Biallelic variants in FLII cause pediatric cardiomyopathy by disrupting cardiomyocyte cell adhesion and myofibril organization

Pediatric cardiomyopathy (CM) represents a group of rare, severe disorders that affect the myocardium. To date, the etiology and mechanisms underlying pediatric CM are incompletely understood, hampering accurate diagnosis and individualized therapy development. Here, we identified biallelic variants in the highly conserved flightless-I (FLII) gene in 3 families with idiopathic, early-onset dilated CM. We demonstrated that patient-specific FLII variants, when brought into the zebrafish genome using CRISPR/Cas9 genome editing, resulted in the manifestation of key aspects of morphological and functional abnormalities of the heart, as observed in our patients. Importantly, using these genetic animal models, complemented with in-depth loss-of-function studies, we provided insights into the function of Flii during ventricular chamber morphogenesis in vivo, including myofibril organization and cardiomyocyte cell adhesion, as well as trabeculation. In addition, we identified Flii function to be important for the regulation of Notch and Hippo signaling, crucial pathways associated with cardiac morphogenesis and function. Taken together, our data provide experimental evidence for a role for FLII in the pathogenesis of pediatric CM and report biallelic variants as a genetic cause of pediatric CM.

Quantitative characterization of zebrafish development based on multiple classifications using Mueller matrix OCT

Organ development analysis plays an important role in assessing an individual' s growth health. In this study, we present a non-invasive method for the quantitative characterization of zebrafish multiple organs during their growth, utilizing Mueller matrix optical coherence tomography (Mueller matrix OCT) in combination with deep learning. Firstly, Mueller matrix OCT was employed to acquire 3D images of zebrafish during development. Subsequently, a deep learning based U-Net network was applied to segment various anatomical structures, including the body, eyes, spine, yolk sac, and swim bladder of the zebrafish. Following segmentation, the volume of each organ was calculated. Finally, the development and proportional trends of zebrafish embryos and organs from day 1 to day 19 were quantitatively analyzed. The obtained quantitative results revealed that the volume development of the fish body and individual organs exhibited a steady growth trend. Additionally, smaller organs, such as the spine and swim bladder, were successfully quantified during the growth process. Our findings demonstrate that the combination of Mueller matrix OCT and deep learning effectively quantify the development of various organs throughout zebrafish embryonic development. This approach offers a more intuitive and efficient monitoring method for clinical medicine and developmental biology studies.

Zebrafish as a Model of Cardiac Pathology and Toxicity: Spotlight on Uremic Toxins

Chronic kidney disease (CKD) is an increasing health care problem. About 10% of the general population is affected by CKD, representing the sixth cause of death in the world. Cardiovascular events are the main mortality cause in CKD, with a cardiovascular risk 10 times higher in these patients than the rate observed in healthy subjects. The gradual decline of the kidney leads to the accumulation of uremic solutes with a negative effect on every organ, especially on the cardiovascular system. Mammalian models, sharing structural and functional similarities with humans, have been widely used to study cardiovascular disease mechanisms and test new therapies, but many of them are rather expensive and difficult to manipulate. Over the last few decades, zebrafish has become a powerful non-mammalian model to study alterations associated with human disease. The high conservation of gene function, low cost, small size, rapid growth, and easiness of genetic manipulation are just some of the features of this experimental model. More specifically, embryonic cardiac development and physiological responses to exposure to numerous toxin substances are similar to those observed in mammals, making zebrafish an ideal model to study cardiac development, toxicity, and cardiovascular disease.

Deep learning is widely applicable to phenotyping embryonic development and disease

Genome editing simplifies the generation of new animal models for congenital disorders. However, the detailed and unbiased phenotypic assessment of altered embryonic development remains a challenge. Here, we explore how deep learning (U-Net) can automate segmentation tasks in various imaging modalities, and we quantify phenotypes of altered renal, neural and craniofacial development in Xenopus embryos in comparison with normal variability. We demonstrate the utility of this approach in embryos with polycystic kidneys (pkd1 and pkd2) and craniofacial dysmorphia (six1). We highlight how in toto light-sheet microscopy facilitates accurate reconstruction of brain and craniofacial structures within X. tropicalis embryos upon dyrk1a and six1 loss of function or treatment with retinoic acid inhibitors. These tools increase the sensitivity and throughput of evaluating developmental malformations caused by chemical or genetic disruption. Furthermore, we provide a library of pre-trained networks and detailed instructions for applying deep learning to the reader's own datasets. We demonstrate the versatility, precision and scalability of deep neural network phenotyping on embryonic disease models. By combining light-sheet microscopy and deep learning, we provide a framework for higher-throughput characterization of embryonic model organisms. This article has an associated 'The people behind the papers' interview.