OrganoID: a versatile deep learning platform for organoid image analysis

Organoids are three-dimensional in vitro tissue models that closely represent the native heterogeneity, microanatomy, and functionality of an organ or diseased tissue. Analysis of organoid morphology, growth, and drug response is challenging due to the diversity in shape and size of organoids, movement through focal planes, and limited options for live-cell staining. Here, we present OrganoID, an open-source image analysis platform that automatically recognizes, labels, and tracks single organoids in brightfield and phase-contrast microscopy. The platform identifies organoid morphology pixel by pixel without the need for fluorescence or transgenic labeling and accurately analyzes a wide range of organoid types in time-lapse microscopy experiments. OrganoID uses a modified u-net neural network with minimal feature depth to encourage model generalization and allow fast execution. The network was trained on images of human pancreatic cancer organoids and was validated on images from pancreatic, lung, colon, and adenoid cystic carcinoma organoids with a mean intersection-over-union of 0.76. OrganoID measurements of organoid count and individual area concurred with manual measurements at 96% and 95% agreement respectively. Tracking accuracy remained above 89% over the duration of a four-day validation experiment. Automated single-organoid morphology analysis of a dose-response experiment identified significantly different organoid circularity after exposure to different concentrations of gemcitabine. The OrganoID platform enables straightforward, detailed, and accurate analysis of organoid images to accelerate the use of organoids as physiologically relevant models in high-throughput research.

Using deep learning to predict the outcome of live birth from more than 10,000 embryo data

Background Recently, the combination of deep learning and time-lapse imaging provides an objective, standard and scientific solution for embryo selection. However, the reported studies were based on blastocyst formation or clinical pregnancy as the end point. To the best of our knowledge, there is no predictive model that uses the outcome of live birth as the predictive end point. Can a deep learning model predict the probability of live birth from time-lapse system? Methods This study retrospectively analyzed the time-lapse data and live birth outcomes of embryos samples from January 2018 to November 2019. We used the SGD optimizer with an initial learning rate of 0.025 and cosine learning rate reduction strategy. The network is randomly initialized and trained for 200 epochs from scratch. The model is quantitively evaluated over a hold-out test and a 5-fold cross-validation by the average area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Results The deep learning model was able to predict live birth outcomes from time-lapse images with an AUC of 0.968 in 5-fold stratified cross-validation. Conclusions This research reported a deep learning model that predicts the live birth outcome of a single blastocyst transfer. This efficient model for predicting the outcome of live births can automatically analyze the time-lapse images of the patient’s embryos without the need for manual embryo annotation and evaluation, and then give a live birth prediction score for each embryo, and sort the embryos by the predicted value.

Effects of a Single Opioid Dose on Gastrointestinal Motility in Rabbits (Oryctolagus cuniculus): Comparisons among Morphine, Butorphanol, and Tramadol

The aim of this study was to evaluate and compare the effects of single doses of butorphanol, morphine, and tramadol on gastrointestinal motility in rabbits (Oryctolagus cuniculus) using non-invasive imaging methods, such as radiographic barium follow through and ultrasonographic contraction counts. Time-lapse radiographic and ultrasound examinations were performed before and after a single intramuscular dose of 5 mg kg−1 butorphanol, 10 mg kg−1 morphine, or 10 mg kg−1 tramadol. Pyloric and duodenal contraction counts by ultrasonography and radiographic repletion scores for the stomach and caecum were analysed using a mixed linear model. No significant effect was noted on ultrasound examinations of pyloric and duodenal contractions after administration of an opioid treatment. Morphine had a significant effect on the stomach and the caecum repletion scores, whereas butorphanol had a significant effect only on the caecum repletion score. Tramadol had no significant effect on the stomach or caecum repletion scores. The present findings suggest that a single dose of 5 mg kg−1 butorphanol or 10 mg kg−1 morphine temporarily slows gastrointestinal transit in healthy rabbits, preventing physiological progression of the alimentary bolus without the induction of ileus. In contrast, a single dose of 10 mg kg−1 tramadol has no such effects.

Computable early Caenorhabditis elegans embryo with a phase field model

Morphogenesis is a precise and robust dynamic process during metazoan embryogenesis, consisting of both cell proliferation and cell migration. Despite the fact that much is known about specific regulations at molecular level, how cell proliferation and migration together drive the morphogenesis at cellular and organismic levels is not well understood. Using Caenorhabditis elegans as the model animal, we present a phase field model to compute early embryonic morphogenesis within a confined eggshell. With physical information about cell division obtained from three-dimensional time-lapse cellular imaging experiments, the model can precisely reproduce the early morphogenesis process as seen in vivo, including time evolution of location and morphology of each cell. Furthermore, the model can be used to reveal key cell-cell attractions critical to the development of C. elegans embryo. Our work demonstrates how genetic programming and physical forces collaborate to drive morphogenesis and provides a predictive model to decipher the underlying mechanism.

Experimental Designs to Study the Aggregation and Colonization of Biofilms by Video Microscopy With Statistical Confidence

The goal of this study was to quantify the variability of confocal laser scanning microscopy (CLSM) time-lapse images of early colonizing biofilms to aid in the design of future imaging experiments. To accomplish this a large imaging dataset consisting of 16 independent CLSM microscopy experiments was leveraged. These experiments were designed to study interactions between human neutrophils and single cells or aggregates of Staphylococcus aureus (S. aureus) during the initial stages of biofilm formation. Results suggest that in untreated control experiments, variability differed substantially between growth phases (i.e., lag or exponential). When studying the effect of an antimicrobial treatment (in this case, neutrophil challenge), regardless of the inoculation level or of growth phase, variability changed as a frown-shaped function of treatment efficacy (i.e., the reduction in biofilm surface coverage). These findings were used to predict the best experimental designs for future imaging studies of early biofilms by considering differing (i) numbers of independent experiments; (ii) numbers of fields of view (FOV) per experiment; and (iii) frame capture rates per hour. A spreadsheet capable of assessing any user-specified design is included that requires the expected mean log reduction and variance components from user-generated experimental results. The methodology outlined in this study can assist researchers in designing their CLSM studies of antimicrobial treatments with a high level of statistical confidence.

A comparative study of intravenous lignocaine, dexamethasone and combination of lignocaine-dexamethasone in attenuating propofol induced pain

Background: General anesthesia is preferred during surgeries to reduce the pain stimuli in patients and to increase the precision of surgical procedure. Propofol is amongst the most widely used general anesthetic agent with limitation of induced pain during administration. Current study was conducted to compare the effect of intravenous pre-administration of various drugs in attenuating propofol induced pain. Methods: A comparative observational study was conducted on patients of either sex and aged between 18-60 years. Patients were divided in three groups, who received intravenous lignocaine, dexamethasone and combination of lignocaine-dexamethasone respectively to attenuate propofol induced pain. Different variables like HR, SBP, DBP, MAP, RR SpO2 and any adverse events were monitored in all the patients. Results: The 46.66% and 53.33% patients who received lignocaine and dexamethasone alone perceived propofol induced mild to moderate pain; while only 23.33% patients who received lignocaine and dexamethasone in combination perceived mild propofol induced pain. The propofol induced pain event was persistent in only 2 out of 30 patients after a time lapse of 30 seconds for the group receiving lignocaine and dexamethasone in combination. Whereas, the pain event was present even after time lapse of 30 seconds in 08 and 07 out of 30 patients of groups receiving lignocaine and dexamethasone alone.Conclusions: Pre-administration of lignocaine and dexamethasone in combination attenuated the propofol induced pain more significantly in comparison to single administration of mentioned drugs. No significant adverse events except perianal irritation were observed in some patients who received combination of lignocaine and dexamethasone intravenously.

NNeurite: artificial neuronal networks for the unsupervised extraction of axonal and dendritic time-lapse signals

Fluorescence microscopy of Ca2+ transients in small neurites of the behaving mouse provides an unprecedented view of the micrometer-scale mechanisms supporting neuronal communication and computation, and therefore opens the way to understanding their role in cognition. However, the exploitation of this growing and precious experimental data is impeded by the scarcity of methods dedicated to the analysis of images of neurites activity in vivo. We present NNeurite, a set of mathematical and computational techniques specialized for the analysis of time-lapse microscopy images of neurite activity in small behaving animals. Starting from noisy and unstable microscopy images containing an unknown number of small neurites, NNeurite simultaneously aligns images, denoises signals and extracts the location and the temporal activity of the sources of Ca2+ transients. At the core of NNeurite is a novel artificial neuronal network(NN) which we have specifically designed to solve the non-negative matrix factorization (NMF)problem modeling source separation in fluorescence microscopy images. For the first time, we have embedded non-rigid image alignment in the NMF optimization procedure, hence allowing to stabilize images based on the transient and weak neurite signals. NNeurite processing is free of any human intervention as NN training is unsupervised and the unknown number of Ca2+ sources is automatically obtained by the NN-based computation of a low-dimensional representation of time-lapse images. Importantly, the spatial shapes of the sources of Ca2+ fluorescence are not constrained in NNeurite, which allowed to automatically extract the micrometer-scale details of dendritic and axonal branches, such dendritic spines and synaptic boutons, in the cortex of behaving mice. We provide NNeurite as a free and open-source library to support the efforts of the community in advancing in vivo microscopy of neurite activity.