Cellular bases of olfactory circuit assembly revealed by systematic time-lapse imaging.

Neural circuit assembly features simultaneous targeting of numerous neuronal processes from constituent neuron types, yet the dynamics is poorly understood. Here, we use the Drosophila olfactory circuit to investigate dynamic cellular processes by which olfactory receptor neurons (ORNs) target axons precisely to specific glomeruli in the ipsi- and contralateral antennal lobes. Time-lapse imaging of individual axons from 30 ORN types revealed a rich diversity in extension speed, innervation timing, and ipsilateral branch locations and identified that ipsilateral targeting occurs via stabilization of transient interstitial branches. Fast imaging using adaptive optics-corrected lattice light-sheet microscopy showed that upon approaching target, many ORN types exhibiting "exploring branches" consisted of parallel microtubule-based terminal branches emanating from an F-actin-rich hub. Antennal nerve ablations uncovered essential roles for bilateral axons in contralateral target selection and for ORN axons to facilitate dendritic refinement of postsynaptic partner neurons. Altogether, these observations provide cellular bases for wiring specificity establishment.

Uptake-independent killing of macrophages by extracellular Mycobacterium tuberculosis aggregates.

Mycobacterium tuberculosis (Mtb) infection is initiated by inhalation of bacteria into lung alveoli, where they are phagocytosed by resident macrophages. Intracellular Mtb replication induces the death of the infected macrophages and the release of bacterial aggregates. Here, we show that these aggregates can evade phagocytosis by killing macrophages in a contact-dependent but uptake-independent manner. We use time-lapse fluorescence microscopy to show that contact with extracellular Mtb aggregates triggers macrophage plasma membrane perturbation, cytosolic calcium accumulation, and pyroptotic cell death. These effects depend on the Mtb ESX-1 secretion system, however, this system alone cannot induce calcium accumulation and macrophage death in the absence of the Mtb surface-exposed lipid phthiocerol dimycocerosate. Unexpectedly, we found that blocking ESX-1-mediated secretion of the EsxA/EsxB virulence factors does not eliminate the uptake-independent killing of macrophages and that the 50-kDa isoform of the ESX-1-secreted protein EspB can mediate killing in the absence of EsxA/EsxB secretion. Treatment with an ESX-1 inhibitor reduces uptake-independent killing of macrophages by Mtb aggregates, suggesting that novel therapies targeting this anti-phagocytic mechanism could prevent the propagation of extracellular bacteria within the lung.

Intercellular contact and cargo transfer between Müller glia and to microglia precede apoptotic cell clearance in the developing retina.

To clarify our understanding of glial phagocytosis in retinal development, we used real-time imaging of larval zebrafish to provide cell-type specific resolution of this process. We show that radial Müller glia frequently participate in microglial phagocytosis while also completing a subset of phagocytic events. Müller glia actively engage with dying cells through initial target cell contact and phagocytic cup formation, after which an exchange of the dying cell from Müller glia to microglia often takes place. In addition, we find evidence that Müller glia cellular material, possibly from the initial Müller cell phagocytic cup, is internalized into microglial compartments. Previously undescribed Müller cell behaviors were seen, including cargo splitting, wrestling for targets and lateral passing of cargo to neighbors. Collectively, our work provides new insight into glial functions and intercellular interactions, which will allow future work to understand these behaviors on a molecular level.

SPIRO - the automated Petri plate imaging platform designed by biologists, for biologists.

Phenotyping of model organisms grown on Petri plates is often carried out manually, despite the procedures being time-consuming and laborious. The main reason for this is the limited availability of automated phenotyping facilities, whereas constructing a custom automated solution can be a daunting task for biologists. Here, we describe SPIRO, the Smart Plate Imaging Robot, an automated platform that acquires time-lapse photographs of up to four vertically oriented Petri plates in a single experiment, corresponding to 192 seedlings for a typical root growth assay and up to 2500 seeds for a germination assay. SPIRO is catered specifically to biologists' needs, requiring no engineering or programming expertise for assembly and operation. Its small footprint is optimized for standard incubators, the inbuilt green LED enables imaging under dark conditions, and remote control provides access to the data without interfering with sample growth. SPIRO's excellent image quality is suitable for automated image processing, which we demonstrate on the example of seed germination and root growth assays. Furthermore, the robot can be easily customized for specific uses, as all information about SPIRO is released under open-source licenses. Importantly, uninterrupted imaging allows considerably more precise assessment of seed germination parameters and root growth rates compared with manual assays. Moreover, SPIRO enables previously technically challenging assays such as phenotyping in the dark. We illustrate the benefits of SPIRO in proof-of-concept experiments which yielded a novel insight on the interplay between autophagy, nitrogen sensing, and photoblastic response.

Genetically Encoded Tools for Optical Dissection of the Mammalian Cell Cycle.

Eukaryotic cells spend most of their life in interphase of the cell cycle. Understanding the rich diversity of metabolic and genomic regulation that occurs in interphase requires the demarcation of precise phase boundaries in situ. Here, we report the properties of two genetically encoded fluorescence sensors, Fucci(CA) and Fucci(SCA), which enable real-time monitoring of interphase and cell-cycle biology. We re-engineered the Cdt1-based sensor from the original Fucci system to respond to S phase-specific CUL4Ddb1-mediated ubiquitylation alone or in combination with SCFSkp2-mediated ubiquitylation. In cultured cells, Fucci(CA) produced a sharp triple color-distinct separation of G1, S, and G2, while Fucci(SCA) permitted a two-color readout of G1 and S/G2. Fucci(CA) applications included tracking the transient G1 phase of rapidly dividing mouse embryonic stem cells and identifying a window for UV-irradiation damage in S phase. These results show that Fucci(CA) is an essential tool for quantitative studies of interphase cell-cycle regulation.

Pancreatic α- and ß-cellular clocks have distinct molecular properties and impact on islet hormone secretion and gene expression.

A critical role of circadian oscillators in orchestrating insulin secretion and islet gene transcription has been demonstrated recently. However, these studies focused on whole islets and did not explore the interplay between α-cell and ß-cell clocks. We performed a parallel analysis of the molecular properties of α-cell and ß-cell oscillators using a mouse model expressing three reporter genes: one labeling α cells, one specific for ß cells, and a third monitoring circadian gene expression. Thus, phase entrainment properties, gene expression, and functional outputs of the α-cell and ß-cell clockworks could be assessed in vivo and in vitro at the population and single-cell level. These experiments showed that α-cellular and ß-cellular clocks are oscillating with distinct phases in vivo and in vitro. Diurnal transcriptome analysis in separated α and ß cells revealed that a high number of genes with key roles in islet physiology, including regulators of glucose sensing and hormone secretion, are differentially expressed in these cell types. Moreover, temporal insulin and glucagon secretion exhibited distinct oscillatory profiles both in vivo and in vitro. Altogether, our data indicate that differential entrainment characteristics of circadian α-cell and ß-cell clocks are an important feature in the temporal coordination of endocrine function and gene expression.

Machine learning in time-lapse imaging to differentiate embryos from young vs old mice†.

Time-lapse microscopy for embryos is a non-invasive technology used to characterize early embryo development. This study employs time-lapse microscopy and machine learning to elucidate changes in embryonic growth kinetics with maternal aging. We analyzed morphokinetic parameters of embryos from young and aged C57BL6/NJ mice via continuous imaging. Our findings show that aged embryos accelerated through cleavage stages (from 5-cells) to morula compared to younger counterparts, with no significant differences observed in later stages of blastulation. Unsupervised machine learning identified two distinct clusters comprising of embryos from aged or young donors. Moreover, in supervised learning, the extreme gradient boosting algorithm successfully predicted the age-related phenotype with 0.78 accuracy, 0.81 precision, and 0.83 recall following hyperparameter tuning. These results highlight two main scientific insights: maternal aging affects embryonic development pace, and artificial intelligence can differentiate between embryos from aged and young maternal mice by a non-invasive approach. Thus, machine learning can be used to identify morphokinetics phenotypes for further studies. This study has potential for future applications in selecting human embryos for embryo transfer, without or in complement with preimplantation genetic testing.

Radial compressed sensing imaging improves the velocity detection limit of single cell tracking time-lapse MRI.

PURPOSE: Time-lapse MRI enables tracking of single iron-labeled cells. Yet, due to temporal blurring, only slowly moving cells can be resolved. To study faster cells for example during inflammatory processes, accelerated acquisition is needed. METHODS: A rotating phantom system was developed to quantitatively measure the current maximum detectable speed of cells in time-lapse MRI. For accelerated cell tracking, an interleaved radial acquisition scheme was applied to phantom and murine brain in vivo time-lapse MRI experiments at 9.4 T. Detection of iron-labeled cells was evaluated in fully sampled and undersampled reconstructions with and without compressed sensing. RESULTS: The rotating phantom system enabled ultra-slow rotation of phantoms and a velocity detection limit of full-brain Cartesian time-lapse MRI of up to 172 µm/min was determined. Both phantom and in vivo measurements showed that single cells can be followed dynamically using radial time-lapse MRI. Higher temporal resolution of undersampled reconstructions reduced geometric distortion, the velocity detection limit was increased to 1.1 mm/min in vitro, and previously hidden fast-moving cells were recovered. In the mouse brain after in vivo labeling, a total of 42 ± 4 cells were counted in fully sampled, but only 7 ± 1 in undersampled images due to streaking artifacts. Using compressed sensing 33 ± 4 cells were detected. CONCLUSION: Interleaved radial time-lapse MRI permits retrospective reconstruction of both fully sampled and accelerated images, enables single cell tracking at higher temporal resolution and recovers cells hidden before due to blurring. The velocity detection limit as determined with the rotating phantom system increased two- to three-fold compared to previous results.

Testing an artificial intelligence algorithm to predict fetal heartbeat of vitrified-warmed blastocysts from a single image: predictive ability in different settings.

STUDY QUESTION: Could an artificial intelligence (AI) algorithm predict fetal heartbeat from images of vitrified-warmed embryos? SUMMARY ANSWER: Applying AI to vitrified-warmed blastocysts may help predict which ones will result in implantation failure early enough to thaw another. WHAT IS KNOWN ALREADY: The application of AI in the field of embryology has already proven effective in assessing the quality of fresh embryos. Therefore, it could also be useful to predict the outcome of frozen embryo transfers, some of which do not recover their pre-vitrification volume, collapse, or degenerate after warming without prior evidence. STUDY DESIGN, SIZE, DURATION: This retrospective cohort study included 1109 embryos from 792 patients. Of these, 568 were vitrified blastocysts cultured in time-lapse systems in the period between warming and transfer, from February 2022 to July 2023. The other 541 were fresh-transferred blastocysts serving as controls. PARTICIPANTS/MATERIALS, SETTING, METHODS: Four types of time-lapse images were collected: last frame of development of 541 fresh-transferred blastocysts (FTi), last frame of 467 blastocysts to be vitrified (PVi), first frame post-warming of 568 vitrified embryos (PW1i), and last frame post-warming of 568 vitrified embryos (PW2i). After providing the images to the AI algorithm, the returned scores were compared with the conventional morphology and fetal heartbeat outcomes of the transferred embryos (n = 1098). The contribution of the AI score to fetal heartbeat was analyzed by multivariate logistic regression in different patient populations, and the predictive ability of the models was measured by calculating the area under the receiver-operating characteristic curve (ROC-AUC). MAIN RESULTS AND THE ROLE OF CHANCE: Fetal heartbeat rate was related to AI score from FTi (P < 0.001), PW1i (P < 0.05), and PW2i (P < 0.001) images. The contribution of AI score to fetal heartbeat was significant in the oocyte donation program for PW2i (odds ratio (OR)=1.13; 95% CI [1.04-1.23]; P < 0.01), and in cycles with autologous oocytes for PW1i (OR = 1.18; 95% CI [1.01-1.38]; P < 0.05) and PW2i (OR = 1.15; 95% CI [1.02-1.30]; P < 0.05), but was not significantly associated with fetal heartbeat in genetically analyzed embryos. AI scores from the four groups of images varied according to morphological category (P < 0.001). The PW2i score differed in collapsed, non-re-expanded, or non-viable embryos compared to normal/viable embryos (P < 0.001). The predictability of the AI score was optimal at a post-warming incubation time of 3.3-4 h (AUC = 0.673). LIMITATIONS, REASONS FOR CAUTION: The algorithm was designed to assess fresh embryos prior to vitrification, but not thawed ones, so this study should be considered an external trial. WIDER IMPLICATIONS OF THE FINDINGS: The application of predictive software in the management of frozen embryo transfers may be a useful tool for embryologists, reducing the cancellation rates of cycles in which the blastocyst does not recover from vitrification. Specifically, the algorithm tested in this research could be used to evaluate thawed embryos both in clinics with time-lapse systems and in those with conventional incubators only, as just a single photo is required. STUDY FUNDING/COMPETING INTERESTS: This study was supported by the Regional Ministry of Innovation, Universities, Science and Digital Society of the Valencian Community (CIACIF/2021/019) and by Instituto de Salud Carlos III (PI21/00283), and co-funded by European Union (ERDF, 'A way to make Europe'). M.M. received personal fees in the last 5 years as honoraria for lectures from Merck, Vitrolife, MSD, Ferring, AIVF, Theramex, Gedeon Richter, Genea Biomedx, and Life Whisperer. There are no other competing interests. TRIAL REGISTRATION NUMBER: N/A.

Time will tell: time-lapse technology and artificial intelligence to set time cut-offs indicating embryo incompetence.

STUDY QUESTION: Can more reliable time cut-offs of embryo developmental incompetence be generated by combining time-lapse technology (TLT), artificial intelligence, and preimplantation genetics screening for aneuploidy (PGT-A)? SUMMARY ANSWER: Embryo developmental incompetence can be better predicted by time cut-offs at multiple developmental stages and for different ranges of maternal age. WHAT IS KNOWN ALREADY: TLT is instrumental for the continual and undisturbed observation of embryo development. It has produced morphokinetic algorithms aimed at selecting embryos able to generate a viable pregnancy, however, such efforts have had limited success. Regardless, the potential of this technology for improving multiple aspects of the IVF process remains considerable. Specifically, TLT could be harnessed to discriminate developmentally incompetent embryos: i.e. those unable to develop to the blastocyst stage or affected by full-chromosome meiotic aneuploidies. If proven valuable, this application would prevent the non-productive use of such embryos, thereby improving laboratory and clinical efficiency and reducing patient stress and costs due to unnecessary embryo transfer and cryopreservation. STUDY DESIGN, SIZE, DURATION: The training dataset involved embryos of PGT-A cycles cultured in Embryoscope with a single media (836 euploid and 1179 aneuploid blastocysts and 1874 arrested embryos; 2013-2020). Selection criteria were ejaculated sperm, own (not donated) fresh oocytes, trophectoderm biopsy and comprehensive-chromosome-testing to diagnose uniform aneuploidies. Out-of-sample (30% of training), internal (299 euploid and 490 aneuploid blastocysts and 680 arrested embryos; 2021-2022) and external (97 euploid, 110 aneuploid and 603 untested blastocysts and 514 arrested embryos, 2018 to early 2022) validations were conducted. PARTICIPANTS/MATERIALS, SETTING, METHODS: A training dataset (70%) was used to define thresholds. Several models were generated by fitting outcomes to each timing (tPNa-t8) and maternal age. ROC curves pinpointed in-sample classification values associated with 95%, 99% and 99.99% true-positive rate for predicting incompetence. These values were integrated with upper limits of maternal age ranges (<35, 35-37, 38-40, 41-42, and >42 years) in logit functions to identify time cut-offs, whose accuracy was tested on the validation datasets through confusion matrices. MAIN RESULTS AND THE ROLE OF CHANCE: For developmental (in)competence, the best performing (i) tPNa cut-offs were 27.8 hpi (error-rate: 0/743), 32.6 hpi (error rate: 0/934), 26.8 hpi (error rate: 0/1178), 22.9 hpi (error-rate: 1/654, 0.1%) and 17.2 hpi (error rate: 4/423, 0.9%) in the <35, 35-37, 38-40, 41-42, and >42 years groups, respectively; (ii) tPNf cut-offs were 36.7 hpi (error rate: 0/738), 47.9 hpi (error rate: 0/921), 45.6 hpi (error rate: 1/1156, 0.1%), 44.1 hpi (error rate: 0/647) and 41.8 hpi (error rate: 0/417); (iii) t2 cut-offs were 50.9 hpi (error rate: 0/724), 49 hpi (error rate: 0/915), 47.1 hpi (error rate: 0/1146), 45.8 hpi (error rate: 0/636) and 43.9 hpi (error rate: 0/416); (iv) t4 cut-offs were 66.9 hpi (error rate: 0/683), 80.7 hpi (error rate: 0/836), 77.1 hpi (error rate: 0/1063), 74.7 hpi (error rate: 0/590) and 71.2 hpi (error rate: 0/389); and (v) t8 cut-offs were 118.1 hpi (error rate: 0/619), 110.6 hpi (error rate: 0/772), 140 hpi (error rate: 0/969), 135 hpi (error rate: 0/533) and 127.5 hpi (error rate: 0/355). tPNf and t2 showed a significant association with chromosomal (in)competence, also when adjusted for maternal age. Nevertheless, the relevant cut-offs were found to perform less well and were redundant compared with the blastocyst development cut-offs. LIMITATIONS, REASONS FOR CAUTION: Study limits are its retrospective design and the datasets being unbalanced towards advanced maternal age cases. The potential effects of abnormal cleavage patterns were not assessed. Larger sample sizes and external validations in other clinical settings are warranted. WIDER IMPLICATIONS OF THE FINDINGS: If confirmed by independent studies, this approach could significantly improve the efficiency of ART, by reducing the workload and patient impacts (extended culture and cleavage stage cryopreservation or transfer) associated with embryos that ultimately are developmentally incompetent and should not be considered for treatment. Pending validation, these data might be applied also in static embryo observation settings. STUDY FUNDING/COMPETING INTEREST(S): This study was supported by the participating institutions. The authors have no conflicts of interest to declare. TRIAL REGISTRATION NUMBER: N/A.

The density of the inner cell mass is a new indicator of the quality of a human blastocyst: a valid supplement to the Gardner scoring system.

STUDY QUESTION: Can the density of the inner cell mass (ICM) be a new indicator of the quality of the human blastocyst? SUMMARY ANSWER: The densification index (DI) developed in this study can quantify ICM density and provide positive guidance for ploidy, pregnancy, and live birth. WHAT IS KNOWN ALREADY: In evaluating the quality of ICM, reproductive care clinics still use size indicators without further evaluation. The main disadvantage of this current method is that the evaluation of blastocyst ICM is relatively rough and cannot meet the needs of clinical embryologists, especially when multiple blastocysts have the same ICM score, which makes them difficult to evaluate further. STUDY DESIGN, SIZE, DURATION: This observational study included data from 2272 blastocysts in 1991 frozen-thawed embryo transfer (FET) cycles between January 2018 to November 2021 and 1105 blastocysts in 430 preimplantation genetic testing cycles between January 2019 and February 2023. PARTICIPANTS/MATERIALS, SETTING, METHODS: FET, ICSI, blastocyst culture, trophectoderm biopsy, time-lapse (TL) monitoring, and next-generation sequencing were performed. After preliminary sample size selection, the 11 focal plane images captured by the TL system were normalized and the spatial frequency was used to construct the DI of the ICM. MAIN RESULTS AND THE ROLE OF CHANCE: This study successfully constructed a quantitative indicator DI that can reflect the degree of ICM density in terms of fusion and texture features. The higher the DI value, the better the density of the blastocyst ICM, and the higher the chances that the blastocyst was euploid (P < 0.001) and that pregnancy (P < 0.001) and live birth (P = 0.005) were reached. In blastocysts with ICM graded B and blastocysts graded 4BB, DI was also positively associated with ploidy, pregnancy, and live birth (P < 0.05). ROC analysis showed that combining the Gardner scoring system with DI can more effectively predict pregnancy and live births, when compared to using the Gardner scoring system alone. LIMITATIONS, REASONS FOR CAUTION: Accurate calculation of the DI value places high demands on image quality, requiring manual selection of the clearest focal plane and exposure control. Images with the ICM not completely within the field of view cannot be used. The association between the density of ICM and chromosomal mosaicism was not evaluated. The associations between the density of ICM and different assisted reproductive technologies and different culture conditions in embryo laboratories were also not evaluated. Prospective studies are needed to further investigate the impact of ICM density on clinical outcomes. WIDER IMPLICATIONS OF THE FINDINGS: ICM density assessment is a new direction in blastocyst assessment. This study explores new ways of assessing blastocyst ICM density and develops quantitative indicators and a corresponding qualitative evaluation scheme for ICM density. The DI of the blastocyst ICM developed in this study is easy to calculate and requires only TL equipment and image processing, providing positive guidance for clinical outcomes. The qualitative evaluation scheme of ICM density can assist embryologists without TL equipment to manually evaluate ICM density. ICM density is a simple indicator that can be used in practice and is a good complement to the blastocyst scoring systems currently used in most centers. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by the National Key Research & Development Program of China (2021YFC2700603). The authors report no financial or commercial conflicts of interest. TRIAL REGISTRATION NUMBER: N/A.

A brief history of artificial intelligence embryo selection: from black-box to glass-box.

With the exponential growth of computing power and accumulation of embryo image data in recent years, artificial intelligence (AI) is starting to be utilized in embryo selection in IVF. Amongst different AI technologies, machine learning (ML) has the potential to reduce operator-related subjectivity in embryo selection while saving labor time on this task. However, as modern deep learning (DL) techniques, a subcategory of ML, are increasingly used, its integrated black-box attracts growing concern owing to the well-recognized issues regarding lack of interpretability. Currently, there is a lack of randomized controlled trials to confirm the effectiveness of such black-box models. Recently, emerging evidence has shown underperformance of black-box models compared to the more interpretable traditional ML models in embryo selection. Meanwhile, glass-box AI, such as interpretable ML, is being increasingly promoted across a wide range of fields and is supported by its ethical advantages and technical feasibility. In this review, we propose a novel classification system for traditional and AI-driven systems from an embryology standpoint, defining different morphology-based selection approaches with an emphasis on subjectivity, explainability, and interpretability.

Cytoplasmic strings in human blastocysts: hypotheses of their role and implications for embryo selection.

STUDY QUESTION: What are the implications of the presence cytoplasmic strings (Cyt-S) and their quantity and dynamics for the pre-implantation development of human blastocysts? SUMMARY ANSWER: Cyt-S are common in human embryos and are associated with faster blastocyst development, larger expansion, and better morphological quality. WHAT IS KNOWN ALREADY: Cyt-S are dynamic cellular projections connecting inner cell mass and trophectoderm (TE) cells, that can be observed during blastocyst expansion. Their prevalence in human embryos has been estimated to be between 44% and 93%. Data relevant to their clinical implications and role in development are lacking, limited, or controversial. STUDY DESIGN, SIZE, DURATION: Retrospective study conducted at a single IVF center between May 2013 and November 2014 and involving 124 pre-implantation genetic testing for aneuploidy cycles in a time-lapse incubator with ≥1 blastocyst biopsied and vitrified (N = 370 embryos assessed). These cycles resulted in 87 vitrified-warmed single-euploid blastocyst transfers. PARTICIPANTS/MATERIALS, SETTING, METHODS: ICSI, continuous blastocyst culture (Days 5-7), TE biopsy of fully expanded blastocysts without Day 3 zona pellucida drilling, qPCR to assess uniform full-chromosome aneuploidies, and vitrification were all performed. Only vitrified-warmed euploid single-embryo-transfers were conducted. Blastocyst morphological quality was defined according to Gardner's criteria. The AI-based software CHLOE™ (Fairtility) automatically registered timings from time of starting blastulation (tSB) to biopsy (t-biopsy, i.e. blastocyst full-expansion) as hours-post-insemination (hpi), embryo area (including zona pellucida in µm2), and spontaneous blastocyst collapses. One senior embryologist manually annotated Cyt-S presence, quantity, timings, and type (thick cell-to-cell connections and/or threads). All significant associations were confirmed through regression analyses. All couples', cycles', and embryos' main features were also tested for associations with Cyt-S presence, quantity, and dynamics. MAIN RESULTS AND THE ROLE OF CHANCE: About 94.3% of the patients (N = 117/124) had ≥1 embryo with Cyt-S. Out of a total of 370 blastocysts, 55 degenerated between blastulation and full-expansion (N = 55/370, 14.9%). The degeneration rate among embryos with ≥1 Cyt-S was 10.8% (N = 33/304), significantly lower than that of embryos without Cyt-S (33.3%, N = 22/66, P < 0.01). Of the remaining 315 viable blastocysts analyzed, 86% (N = 271/315; P < 0.01) had ≥1 Cyt-S, on average 3.5 ± 2.1 per embryo ranging 1-13. The first Cyt-S per viable embryo appeared at 115.3 ± 12.5 hpi (85.7-157.7), corresponding to 10.5 ± 5.8 h (0.5-31) after tSB. Overall, we analyzed 937 Cyt-S showing a mean duration of 3.8 ± 2.7 h (0.3-20.9). Cyt-S were mostly threads (N = 508/937, 54.2%) or thick cell-to-cell connections becoming threads (N = 382/937, 40.8%) than thick bridges (N = 47/937, 5.0%). The presence and quantity of Cyt-S were significantly associated with developmentally faster (on average 6-12 h faster) and more expanded (on average 2700 µm2-larger blastocyst's area at t-biopsy) embryos. Also, the presence and duration of Cyt-S were associated with better morphology. Lastly, while euploidy rates were comparable between blastocysts with and without Cyt-S, all euploid blastocysts transferred from the latter group failed to implant (N = 10). LIMITATIONS, REASONS FOR CAUTION: Cyt-S presence and dynamics were assessed manually on seven focal planes from video frames recorded every 15 min. The patients included were mostly of advanced maternal age. Only associations could be reported, but no causations/consequences. Lastly, larger datasets are required to better assess Cyt-S associations with clinical outcomes. WIDER IMPLICATIONS OF THE FINDINGS: Cyt-S are common during human blastocyst expansion, suggesting their physiological implication in this process. Their presence, quantity and dynamics mirror embryo viability, and morphological quality, yet their role is still unknown. Future basic science studies are encouraged to finally describe Cyt-S molecular nature and biophysical properties, and Artificial Intelligence tools should aid these studies by incorporating Cyt-S assessment. STUDY FUNDING/COMPETING INTEREST(S): None. TRIAL REGISTRATION NUMBER: N/A.

Generative artificial intelligence to produce high-fidelity blastocyst-stage embryo images.

STUDY QUESTION: Can generative artificial intelligence (AI) models produce high-fidelity images of human blastocysts? SUMMARY ANSWER: Generative AI models exhibit the capability to generate high-fidelity human blastocyst images, thereby providing substantial training datasets crucial for the development of robust AI models. WHAT IS KNOWN ALREADY: The integration of AI into IVF procedures holds the potential to enhance objectivity and automate embryo selection for transfer. However, the effectiveness of AI is limited by data scarcity and ethical concerns related to patient data privacy. Generative adversarial networks (GAN) have emerged as a promising approach to alleviate data limitations by generating synthetic data that closely approximate real images. STUDY DESIGN, SIZE, DURATION: Blastocyst images were included as training data from a public dataset of time-lapse microscopy (TLM) videos (n = 136). A style-based GAN was fine-tuned as the generative model. PARTICIPANTS/MATERIALS, SETTING, METHODS: We curated a total of 972 blastocyst images as training data, where frames were captured within the time window of 110-120 h post-insemination at 1-h intervals from TLM videos. We configured the style-based GAN model with data augmentation (AUG) and pretrained weights (Pretrained-T: with translation equivariance; Pretrained-R: with translation and rotation equivariance) to compare their optimization on image synthesis. We then applied quantitative metrics including Fréchet Inception Distance (FID) and Kernel Inception Distance (KID) to assess the quality and fidelity of the generated images. Subsequently, we evaluated qualitative performance by measuring the intelligence behavior of the model through the visual Turing test. To this end, 60 individuals with diverse backgrounds and expertise in clinical embryology and IVF evaluated the quality of synthetic embryo images. MAIN RESULTS AND THE ROLE OF CHANCE: During the training process, we observed consistent improvement of image quality that was measured by FID and KID scores. Pretrained and AUG + Pretrained initiated with remarkably lower FID and KID values compared to both Baseline and AUG + Baseline models. Following 5000 training iterations, the AUG + Pretrained-R model showed the highest performance of the evaluated five configurations with FID and KID scores of 15.2 and 0.004, respectively. Subsequently, we carried out the visual Turing test, such that IVF embryologists, IVF laboratory technicians, and non-experts evaluated the synthetic blastocyst-stage embryo images and obtained similar performance in specificity with marginal differences in accuracy and sensitivity. LIMITATIONS, REASONS FOR CAUTION: In this study, we primarily focused the training data on blastocyst images as IVF embryos are primarily assessed in blastocyst stage. However, generation of an array of images in different preimplantation stages offers further insights into the development of preimplantation embryos and IVF success. In addition, we resized training images to a resolution of 256 × 256 pixels to moderate the computational costs of training the style-based GAN models. Further research is needed to involve a more extensive and diverse dataset from the formation of the zygote to the blastocyst stage, e.g. video generation, and the use of improved image resolution to facilitate the development of comprehensive AI algorithms and to produce higher-quality images. WIDER IMPLICATIONS OF THE FINDINGS: Generative AI models hold promising potential in generating high-fidelity human blastocyst images, which allows the development of robust AI models as it can provide sufficient training datasets while safeguarding patient data privacy. Additionally, this may help to produce sufficient embryo imaging training data with different (rare) abnormal features, such as embryonic arrest, tripolar cell division to avoid class imbalances and reach to even datasets. Thus, generative models may offer a compelling opportunity to transform embryo selection procedures and substantially enhance IVF outcomes. STUDY FUNDING/COMPETING INTEREST(S): This study was supported by a Horizon 2020 innovation grant (ERIN, grant no. EU952516) and a Horizon Europe grant (NESTOR, grant no. 101120075) of the European Commission to A.S. and M.Z.E., the Estonian Research Council (grant no. PRG1076) to A.S., and the EVA (Erfelijkheid Voortplanting & Aanleg) specialty program (grant no. KP111513) of Maastricht University Medical Centre (MUMC+) to M.Z.E. TRIAL REGISTRATION NUMBER: Not applicable.

A comparison of morphokinetic models and morphological selection for prioritizing euploid embryos: a multicentre cohort study.

STUDY QUESTION: Are morphokinetic models better at prioritizing a euploid embryo for transfer over morphological selection by an embryologist? SUMMARY ANSWER: Morphokinetic algorithms lead to an improved prioritization of euploid embryos when compared to embryologist selection. WHAT IS KNOWN ALREADY: PREFER (predicting euploidy for embryos in reproductive medicine) is a previously published morphokinetic model associated with live birth and miscarriage. The second model uses live birth as the target outcome (LB model). STUDY DESIGN, SIZE, DURATION: Data for this cohort study were obtained from 1958 biopsied blastocysts at nine IVF clinics across the UK from January 2021 to December 2022. PARTICIPANTS/MATERIALS, SETTING, METHODS: The ability of the PREFER and LB models to prioritize a euploid embryo was compared against arbitrary selection and the prediction of four embryologists using the timelapse video, blinded to the morphokinetic time stamp. The comparisons were made using calculated percentages and normalized discounted cumulative gain (NDCG), whereby an NDCG score of 1 would equate to all euploid embryos being ranked first. In arbitrary selection, the ploidy status was randomly assigned within each cycle and the NDGC calculated, and this was then repeated 100 times and the mean obtained. MAIN RESULTS AND THE ROLE OF CHANCE: Arbitrary embryo selection would rank a euploid embryo first 37% of the time, embryologist selection 39%, and the LB and PREFER ploidy morphokinetic models 46% and 47% of the time, respectively. The AUC for LB and PREFER model was 0.62 and 0.63, respectively. Morphological selection did not significantly improve the performance of both morphokinetic models when used in combination. There was a significant difference between the NDGC metric of the PREFER model versus embryologist selection at 0.96 and 0.87, respectively (t = 14.1, P < 0.001). Similarly, there was a significant difference between the LB model and embryologist selection with an NDGC metric of 0.95 and 0.87, respectively (t = 12.0, P < 0.001). All four embryologists ranked embryos similarly, with an intraclass coefficient of 0.91 (95% CI 0.82-0.95, P < 0.001). LIMITATIONS, REASONS FOR CAUTION: Aside from the retrospective study design, limitations include allowing the embryologist to watch the time lapse video, potentially providing more information than a truly static morphological assessment. Furthermore, the embryologists at the participating centres were familiar with the significant variables in time lapse, which could bias the results. WIDER IMPLICATIONS OF THE FINDINGS: The present study shows that the use of morphokinetic models, namely PREFER and LB, translates into improved euploid embryo selection. STUDY FUNDING/COMPETING INTEREST(S): This study received no specific grant funding from any funding agency in the public, commercial or not-for-profit sectors. Dr Alison Campbell is minor share holder of Care Fertility. All other authors have no conflicts of interest to declare. Time lapse is a technology for which patients are charged extra at participating centres. TRIAL REGISTRATION NUMBER: N/A.

Deep learning to extract the meteorological by-catch of wildlife cameras.

Microclimate-proximal climatic variation at scales of metres and minutes-can exacerbate or mitigate the impacts of climate change on biodiversity. However, most microclimate studies are temperature centric, and do not consider meteorological factors such as sunshine, hail and snow. Meanwhile, remote cameras have become a primary tool to monitor wild plants and animals, even at micro-scales, and deep learning tools rapidly convert images into ecological data. However, deep learning applications for wildlife imagery have focused exclusively on living subjects. Here, we identify an overlooked opportunity to extract latent, ecologically relevant meteorological information. We produce an annotated image dataset of micrometeorological conditions across 49 wildlife cameras in South Africa's Maloti-Drakensberg and the Swiss Alps. We train ensemble deep learning models to classify conditions as overcast, sunshine, hail or snow. We achieve 91.7% accuracy on test cameras not seen during training. Furthermore, we show how effective accuracy is raised to 96% by disregarding 14.1% of classifications where ensemble member models did not reach a consensus. For two-class weather classification (overcast vs. sunshine) in a novel location in Svalbard, Norway, we achieve 79.3% accuracy (93.9% consensus accuracy), outperforming a benchmark model from the computer vision literature (75.5% accuracy). Our model rapidly classifies sunshine, snow and hail in almost 2 million unlabelled images. Resulting micrometeorological data illustrated common seasonal patterns of summer hailstorms and autumn snowfalls across mountains in the northern and southern hemispheres. However, daily patterns of sunshine and shade diverged between sites, impacting daily temperature cycles. Crucially, we leverage micrometeorological data to demonstrate that (1) experimental warming using open-top chambers shortens early snow events in autumn, and (2) image-derived sunshine marginally outperforms sensor-derived temperature when predicting bumblebee foraging. These methods generate novel micrometeorological variables in synchrony with biological recordings, enabling new insights from an increasingly global network of wildlife cameras.

Multiple collapses of blastocysts after full blastocyst formation is an independent risk factor for aneuploidy - a study based on AI and manual validation.

BACKGROUND: The occurrence of blastocyst collapse may become an indicator of preimplantation embryo quality assessment. It has been reported that collapsing blastocysts can lead to higher rates of aneuploidy and poorer clinical outcomes, but more large-scale studies are needed to explore this relationship. This study explored the characteristics of blastocyst collapse identified and quantified by artificial intelligence and explored the associations between blastocyst collapse and embryo ploidy, morphological quality, and clinical outcomes. METHODS: This observational study included data from 3288 biopsied blastocysts in 1071 time-lapse preimplantation genetic testing cycles performed between January 2019 and February 2023 at a single academic fertility center. All transferred blastocysts are euploid blastocysts. The artificial intelligence recognized blastocyst collapse in time-lapse microscopy videos and then registered the collapsing times, and the start time, the recovery duration, the shrinkage percentage of each collapse. The effects of blastocyst collapse and embryo ploidy, pregnancy, live birth, miscarriage, and embryo quality were studied using available data from 1196 euploid embryos and 1300 aneuploid embryos. RESULTS: 5.6% of blastocysts collapsed at least once only before the full blastocyst formation (tB), 19.4% collapsed at least once only after tB, and 3.1% collapsed both before and after tB. Multiple collapses of blastocysts after tB (times ≥ 2) are associated with higher aneuploid rates (54.6%, P > 0.05; 70.5%, P < 0.001; 72.5%, P = 0.004; and 71.4%, P = 0.049 in blastocysts collapsed 1, 2, 3 or ≥ 4 times), which remained significant after adjustment for confounders (OR = 2.597, 95% CI 1.464-4.607, P = 0.001). Analysis of the aneuploid embryos showed a higher ratio of collapses and multiple collapses after tB in monosomies and embryos with subchromosomal deletion of segmental nature (P < 0.001). Blastocyst collapse was associated with delayed embryonic development and declined blastocyst quality. There is no significant difference in pregnancy and live birth rates between collapsing and non-collapsing blastocysts. CONCLUSIONS: Blastocyst collapse is common during blastocyst development. This study underlined that multiple blastocyst collapses after tB may be an independent risk factor for aneuploidy which should be taken into account by clinicians and embryologists when selecting blastocysts for transfer.

Enhancing clinical utility: deep learning-based embryo scoring model for non-invasive aneuploidy prediction.

BACKGROUND: The best method for selecting embryos ploidy is preimplantation genetic testing for aneuploidies (PGT-A). However, it takes more labour, money, and experience. As such, more approachable, non- invasive techniques were still needed. Analyses driven by artificial intelligence have been presented recently to automate and objectify picture assessments. METHODS: In present retrospective study, a total of 3448 biopsied blastocysts from 979 Time-lapse (TL)-PGT cycles were retrospectively analyzed. The "intelligent data analysis (iDA) Score" as a deep learning algorithm was used in TL incubators and assigned each blastocyst with a score between 1.0 and 9.9. RESULTS: Significant differences were observed in iDAScore among blastocysts with different ploidy. Additionally, multivariate logistic regression analysis showed that higher scores were significantly correlated with euploidy (p < 0.001). The Area Under the Curve (AUC) of iDAScore alone for predicting euploidy embryo is 0.612, but rose to 0.688 by adding clinical and embryonic characteristics. CONCLUSIONS: This study provided additional information to strengthen the clinical applicability of iDAScore. This may provide a non-invasive and inexpensive alternative for patients who have no available blastocyst for biopsy or who are economically disadvantaged. However, the accuracy of embryo ploidy is still dependent on the results of next-generation sequencing technology (NGS) analysis.

Investigating developmental characteristics of biopsied blastocysts stratified by mitochondrial copy numbers using time-lapse monitoring.

BACKGROUND: For in vitro fertilization (IVF), mitochondrial DNA (mtDNA) levels in the trophectodermal (TE) cells of biopsied blastocysts have been suggested to be associated with the cells' developmental potential. However, scholars have reached differing opinions regarding the use of mtDNA levels as a reliable biomarker for predicting IVF outcomes. Therefore, this study aims to assess the association of mitochondrial copy number measured by mitoscore associated with embryonic developmental characteristics and ploidy. METHODS: This retrospective study analyzed the developmental characteristics of embryos and mtDNA levels in biopsied trophectodermal cells. The analysis was carried out using time-lapse monitoring and next-generation sequencing from September 2021 to September 2022. Five hundred and fifteen blastocysts were biopsied from 88 patients undergoing IVF who met the inclusion criteria. Embryonic morphokinetics and morphology were evaluated at 118 h after insemination using all recorded images. Blastocysts with appropriate morphology on day 5 or 6 underwent TE biopsy and preimplantation genetic testing for aneuploidy (PGT-A). Statistical analysis involved generalized estimating equations, Pearson's chi-squared test, Fisher's exact test, and Kruskal-Wallis test, with a significance level set at P < 0.05. RESULTS: To examine differences in embryonic characteristics between blastocysts with low versus high mitoscores, the blastocysts were divided into quartiles based on their mitoscore. Regarding morphokinetic characteristics, no significant differences in most developmental kinetics and observed cleavage dysmorphisms were discovered. However, blastocysts in mitoscore group 1 had a longer time for reaching 3-cell stage after tPNf (t3; median: 14.4 h) than did those in mitoscore group 2 (median: 13.8 h) and a longer second cell cycle (CC2; median: 11.7 h) than did blastocysts in mitoscore groups 2 (median: 11.3 h) and 4 (median: 11.4 h; P < 0.05). Moreover, blastocysts in mitoscore group 4 had a lower euploid rate (22.6%) and a higher aneuploid rate (59.1%) than did those in the other mitoscore groups (39.6-49.3% and 30.3-43.2%; P < 0.05). The rate of whole-chromosomal alterations in mitoscore group 4 (63.4%) was higher than that in mitoscore groups 1 (47.3%) and 2 (40.1%; P < 0.05). A multivariate logistic regression model was used to analyze associations between the mitoscore and euploidy of elective blastocysts. After accounting for factors that could potentially affect the outcome, the mitoscore still exhibited a negative association with the likelihood of euploidy (adjusted OR = 0.581, 95% CI: 0.396-0.854; P = 0.006). CONCLUSIONS: Blastocysts with varying levels of mitochondrial DNA, identified through biopsies, displayed similar characteristics in their early preimplantation development as observed through time-lapse imaging. However, the mitochondrial DNA level determined by the mitoscore can be used as a standalone predictor of euploidy.

Assessment of artificial intelligence model and manual morphokinetic annotation system as embryo grading methods for successful live birth prediction: a retrospective monocentric study.

PURPOSE: The introduction of the time-lapse monitoring system (TMS) and the development of predictive algorithms could contribute to the optimal embryos selection for transfer. Therefore, the present study aims at investigating the efficiency of KIDScore and iDAScore systems for blastocyst stage embryos in predicting live birth events. METHODS: The present retrospective study was conducted in a private IVF Unit setting throughout a 10-month period from October 2021 to July 2022, and included the analysis of 429 embryos deriving from 91 IVF/ICSI cycles conducted due to infertility of various etiologies. Embryos incubated at the Embryoscope+ timelapse incubator were analyzed through the established scoring systems: KIDScore and iDAScore®. The main outcome measure was the comparison of the two scoring systems in terms of live birth prediction. Embryos with the higher scores at day 5 (KID5 score/iDA5 score) were transferred or cryopreserved for later use. RESULTS: Embryos with high KID5 and iDA5 scores positively correlated with the probability of successful live birth, with KID5 score yielding a higher efficiency in predicting a successful reproductive outcome compared to a proportionally high iDA5 score. KID5 demonstrated conservative performance in successfully predicting live birth compared to iDA5 score, indicating that an efficient prediction can be either provided by a relatively lower KID5 score or a relatively higher iDA5 score. CONCLUSION: The developed artificial intelligence tools should be implemented in clinical practice in conjunction with the conventional morphological assessment for the conduction of optimized embryo transfer in terms of a successful live birth.