Deep Learning Neural Networks Highly Predict Very Early Onset of Pluripotent Stem Cell Differentiation

Waisman, Ariel; Greca, Alejandro La; Möbbs, Alan Miqueas; Scarafía, María Agustina; Velazque, Natalia Lucía Santín; Neiman, Gabriel; Moro, Lucía; Luzzani, Carlos; Sevlever, Gustavo; Guberman, Alejandra; Miriuka, Santiago Gabriel

doi:10.1016/j.stemcr.2019.02.004

Cited by 102 publications

(87 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To date, nearly all cell biological 615 applications of this technology have used some form of supervised learning for 616 classification tasks (44). For example, CNNs have been effectively used for cell-by-cell 617 classification with goals including predicting differentiation of stem cells (45), 618 classifying cell cycle status (46), and identifying cell types based on their motility 619 behavior (47). Pixel-by-pixel classification strategies for cell images have also been 620 fruitfully applied to the problem of image segmentation, using human-annotated "ground 621 truth" images for training purposes (48).…”

Section: Recent Advances In the Design Of Deep Convolutional Neural Nmentioning

confidence: 99%

Quantitative comparison of principal component analysis and unsupervised deep learning using variational autoencoders for shape analysis of motile cells

Chan

Hadjitheodorou

Tsai

et al. 2020

Preprint

View full text Add to dashboard Cite

Cell motility is a crucial biological function for many cell types, including the immune cells in our body that act as first responders to foreign agents. In this work we consider the amoeboid motility of human neutrophils, which show complex and continuous morphological changes during locomotion. We imaged live neutrophils migrating on a 2D plane and extracted unbiased shape representations using cell contours and binary masks. We were able to decompose these complex shapes into low-dimensional encodings with both principal component analysis (PCA) and an unsupervised deep learning technique using variational autoencoders (VAE), enhanced with generative adversarial networks (GANs). We found that the neural network architecture, the VAE-GAN, was able to encode complex cell shapes into a low-dimensional latent space that encodes the same shape variation information as PCA, but much more efficiently. Contrary to the conventional viewpoint that the latent space is a “black box”, we demonstrated that the information learned and encoded within the latent space is consistent with PCA and is reproducible across independent training runs. Furthermore, by including cell speed into the training of the VAE-GAN, we were able to incorporate cell shape and speed into the same latent space. Our work provides a quantitative framework that connects biological form, through cell shape, to a biological function, cell movement. We believe that our quantitative approach to calculating a compact representation of cell shape using the VAE-GAN provides an important avenue that will support further mechanistic dissection of cell motility.AUTHOR SUMMARYDeep convolutional neural networks have recently enjoyed a surge in popularity, and have found useful applications in many fields, including biology. Supervised deep learning, which involves the training of neural networks using existing labeled data, has been especially popular in solving image classification problems. However, biological data is often highly complex and continuous in nature, where prior labeling is impractical, if not impossible. Unsupervised deep learning promises to discover trends in the data by reducing its complexity while retaining the most relevant information. At present, challenges in the extraction of meaningful human-interpretable information from the neural network’s nonlinear discovery process have earned it a reputation of being a “black box” that can perform impressively well at prediction but cannot be used to shed any meaningful insight on underlying mechanisms of variation in biological data sets. Our goal in this paper is to establish unsupervised deep learning as a practical tool to gain scientific insight into biological data by first establishing the interpretability of our particular data set (images of the shapes of motile neutrophils) using more traditional techniques. Using the insight gained from this as a guide allows us to shine light into the “black box” of unsupervised deep learning.

show abstract

Section: Recent Advances In the Design Of Deep Convolutional Neural Nmentioning

confidence: 99%

Quantitative comparison of principal component analysis and unsupervised deep learning using variational autoencoders for shape analysis of motile cells

Chan

Hadjitheodorou

Tsai

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In particular, the use of these technologies in automation may change the way readings are performed. We have already previously shown that this may be the case using a stem cell differentiation model (10 ). Hence, deep learning algorithms may substitute everyday assays in some circumstances.…”

Section: Discussionmentioning

confidence: 96%

“…Therefore, one of the most active field is image recognition (8 , 9 ). We have recently published that NN can be used to classify transmitted light microscopy (TLM) images (10 ). We were able to correctly classify pluripotent stem cell differentiation at one hour or even less, with an accuracy higher than 99%.…”

Section: Introductionmentioning

confidence: 95%

celldeath: a tool for detection of cell death in transmitted light microscopy images by deep learning-based visual recognition

Greca

Pérez

Milone

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Cell death experiments are routinely done in many labs around the world, these experiments are the backbone of many assays for drug development. Cell death detection is usually performed in many ways, and requires time and reagents. However, cell death is preceded by slight morphological changes in cell shape and texture. In this paper, we trained a neural network to classify cells undergoing cell death. We found that the network was able to highly predict cell death after one hour of exposure to camptothecin. Moreover, this prediction largely outperforms human ability. Finally, we provide a simple python tool that can broadly be used to detect cell death.

show abstract

“…In contrast to other established methods such as immunostaining, gene expression analysis, and flow cytometry, cell lines expressing fluorescently tagged proteins offer advantages by enabling non-destructive and minimally invasive assays for tracking the status of cellular function. Deep learning software can be particularly advantageous when analyzing subtle changes in cellular phenotypes from cell imaging data, either by monitoring cultures of iPSCs and early-stage differentiation, 111 determining maturation, or assaying drug responses using experimental cultures. Such methods have recently been validated to determine the quality of iPSC cultures, 112 detect early differentiation changes, 111 and for late-stage analysis such as the categorization of drug responses using iPSC-cardiomyocytes.…”

Section: Cardiomyocytesmentioning

confidence: 99%

“…Deep learning software can be particularly advantageous when analyzing subtle changes in cellular phenotypes from cell imaging data, either by monitoring cultures of iPSCs and early-stage differentiation, 111 determining maturation, or assaying drug responses using experimental cultures. Such methods have recently been validated to determine the quality of iPSC cultures, 112 detect early differentiation changes, 111 and for late-stage analysis such as the categorization of drug responses using iPSC-cardiomyocytes. 113 To generate imaging data with live cells, many engineered reporter iPSC-lines are being developed to aid in the determination of cellular structure and function, including fluorescent reporter lines for structural features 114 and a GCaMP calcium sensor.…”

Section: Cardiomyocytesmentioning

confidence: 99%

Microengineered systems with iPSC-derived cardiac and hepatic cells to evaluate drug adverse effects

Dame

Ribeiro

2020

Exp Biol Med (Maywood)

View full text Add to dashboard Cite

Hepatic and cardiac drug adverse effects are among the leading causes of attrition in drug development programs, in part due to predictive failures of current animal or in vitro models. Hepatocytes and cardiomyocytes differentiated from human induced pluripotent stem cells (iPSCs) hold promise for predicting clinical drug effects, given their human-specific properties and their ability to harbor genetically determined characteristics that underlie inter-individual variations in drug response. Currently, the fetal-like properties and heterogeneity of hepatocytes and cardiomyocytes differentiated from iPSCs make them physiologically different from their counterparts isolated from primary tissues and limit their use for predicting clinical drug effects. To address this hurdle, there have been ongoing advances in differentiation and maturation protocols to improve the quality and use of iPSC-differentiated lineages. Among these are in vitro hepatic and cardiac cellular microsystems that can further enhance the physiology of cultured cells, can be used to better predict drug adverse effects, and investigate drug metabolism, pharmacokinetics, and pharmacodynamics to facilitate successful drug development. In this article, we discuss how cellular microsystems can establish microenvironments for these applications and propose how they could be used for potentially controlling the differentiation of hepatocytes or cardiomyocytes. The physiological relevance of cells is enhanced in cellular microsystems by simulating properties of tissue microenvironments, such as structural dimensionality, media flow, microfluidic control of media composition, and co-cultures with interacting cell types. Recent studies demonstrated that these properties also affect iPSC differentiations and we further elaborate on how they could control differentiation efficiency in microengineered devices. In summary, we describe recent advances in the field of cellular microsystems that can control the differentiation and maturation of hepatocytes and cardiomyocytes for drug evaluation. We also propose how future research with iPSCs within engineered microenvironments could enable their differentiation for scalable evaluations of drug effects. Impact statement Cardiac and hepatic adverse drug effects are among the leading causes of attrition in preclinical and clinical drug development programs as well as marketing withdrawals. The insufficiency of animal testing models has led to considerable interest in the employment of cardiac and hepatic models using human-induced pluripotent stem cells (iPSCs) for drug toxicity testing. However, current batches of iPSC-derived cardiomyocytes and hepatocytes are variable and not matured as adult primary tissues, which limit their prediction of drug effects. This article discusses how the use of microfluidics can create microenvironments to better control differentiation protocols and increase the physiological relevance of iPSC-derived cardiomyocytes and hepatocytes. Development and standardization of technologies will enable evaluation of the potential value of cellular microsystems to improve the in vitro models used in drug development programs. Future steps in this field include controlled connections of organ systems to better recreate clinical metabolism and pharmacokinetics.

show abstract

Deep Learning Neural Networks Highly Predict Very Early Onset of Pluripotent Stem Cell Differentiation

Cited by 102 publications

References 26 publications

Quantitative comparison of principal component analysis and unsupervised deep learning using variational autoencoders for shape analysis of motile cells

Quantitative comparison of principal component analysis and unsupervised deep learning using variational autoencoders for shape analysis of motile cells

celldeath: a tool for detection of cell death in transmitted light microscopy images by deep learning-based visual recognition

Microengineered systems with iPSC-derived cardiac and hepatic cells to evaluate drug adverse effects

Contact Info

Product

Resources

About