BlastAssist: a deep learning pipeline to measure interpretable features of human embryos.

STUDY QUESTION: Can the BlastAssist deep learning pipeline perform comparably to or outperform human experts and embryologists at measuring interpretable, clinically relevant features of human embryos in IVF? SUMMARY ANSWER: The BlastAssist pipeline can measure a comprehensive set of interpretable features of human embryos and either outperform or perform comparably to embryologists and human experts in measuring these features. WHAT IS KNOWN ALREADY: Some studies have applied deep learning and developed 'black-box' algorithms to predict embryo viability directly from microscope images and videos but these lack interpretability and generalizability. Other studies have developed deep learning networks to measure individual features of embryos but fail to conduct careful comparisons to embryologists' performance, which are fundamental to demonstrate the network's effectiveness. STUDY DESIGN, SIZE, DURATION: We applied the BlastAssist pipeline to 67 043 973 images (32 939 embryos) recorded in the IVF lab from 2012 to 2017 in Tel Aviv Sourasky Medical Center. We first compared the pipeline measurements of individual images/embryos to manual measurements by human experts for sets of features, including: (i) fertilization status (n = 207 embryos), (ii) cell symmetry (n = 109 embryos), (iii) degree of fragmentation (n = 6664 images), and (iv) developmental timing (n = 21 036 images). We then conducted detailed comparisons between pipeline outputs and annotations made by embryologists during routine treatments for features, including: (i) fertilization status (n = 18 922 embryos), (ii) pronuclei (PN) fade time (n = 13 781 embryos), (iii) degree of fragmentation on Day 2 (n = 11 582 embryos), and (iv) time of blastulation (n = 3266 embryos). In addition, we compared the pipeline outputs to the implantation results of 723 single embryo transfer (SET) cycles, and to the live birth results of 3421 embryos transferred in 1801 cycles. PARTICIPANTS/MATERIALS, SETTING, METHODS: In addition to EmbryoScope™ image data, manual embryo grading and annotations, and electronic health record (EHR) data on treatment outcomes were also included. We integrated the deep learning networks we developed for individual features to construct the BlastAssist pipeline. Pearson's χ2 test was used to evaluate the statistical independence of individual features and implantation success. Bayesian statistics was used to evaluate the association of the probability of an embryo resulting in live birth to BlastAssist inputs. MAIN RESULTS AND THE ROLE OF CHANCE: The BlastAssist pipeline integrates five deep learning networks and measures comprehensive, interpretable, and quantitative features in clinical IVF. The pipeline performs similarly or better than manual measurements. For fertilization status, the network performs with very good parameters of specificity and sensitivity (area under the receiver operating characteristics (AUROC) 0.84-0.94). For symmetry score, the pipeline performs comparably to the human expert at both 2-cell (r = 0.71 ± 0.06) and 4-cell stages (r = 0.77 ± 0.07). For degree of fragmentation, the pipeline (acc = 69.4%) slightly under-performs compared to human experts (acc = 73.8%). For developmental timing, the pipeline (acc = 90.0%) performs similarly to human experts (acc = 91.4%). There is also strong agreement between pipeline outputs and annotations made by embryologists during routine treatments. For fertilization status, the pipeline and embryologists strongly agree (acc = 79.6%), and there is strong correlation between the two measurements (r = 0.683). For degree of fragmentation, the pipeline and embryologists mostly agree (acc = 55.4%), and there is also strong correlation between the two measurements (r = 0.648). For both PN fade time (r = 0.787) and time of blastulation (r = 0.887), there's strong correlation between the pipeline and embryologists. For SET cycles, 2-cell time (P < 0.01) and 2-cell symmetry (P < 0.03) are significantly correlated with implantation success rate, while other features showed correlations with implantation success without statistical significance. In addition, 2-cell time (P < 5 × 10-11), PN fade time (P < 5 × 10-10), degree of fragmentation on Day 3 (P < 5 × 10-4), and 2-cell symmetry (P < 5 × 10-3) showed statistically significant correlation with the probability of the transferred embryo resulting in live birth. LIMITATIONS, REASONS FOR CAUTION: We have not tested the BlastAssist pipeline on data from other clinics or other time-lapse microscopy (TLM) systems. The association study we conducted with live birth results do not take into account confounding variables, which will be necessary to construct an embryo selection algorithm. Randomized controlled trials (RCT) will be necessary to determine whether the pipeline can improve success rates in clinical IVF. WIDER IMPLICATIONS OF THE FINDINGS: BlastAssist provides a comprehensive and holistic means of evaluating human embryos. Instead of using a black-box algorithm, BlastAssist outputs meaningful measurements of embryos that can be interpreted and corroborated by embryologists, which is crucial in clinical decision making. Furthermore, the unprecedentedly large dataset generated by BlastAssist measurements can be used as a powerful resource for further research in human embryology and IVF. STUDY FUNDING/COMPETING INTEREST(S): This work was supported by Harvard Quantitative Biology Initiative, the NSF-Simons Center for Mathematical and Statistical Analysis of Biology at Harvard (award number 1764269), the National Institute of Heath (award number R01HD104969), the Perelson Fund, and the Sagol fund for embryos and stem cells as part of the Sagol Network. The authors declare no competing interests. TRIAL REGISTRATION NUMBER: Not applicable.

Treatment with Laevo (L)-carnitine reverses the mitochondrial function of human embryos.

PURPOSE: Laevo (l)-carnitine plays important roles in reducing the cytotoxic effects of free fatty acids by forming acyl-carnitine and promoting beta-oxidation, leading to alleviation of cell damage. Recently, the mitochondrial functions in morula has been shown to decrease with the maternal age. Here, we assessed the effect of l-carnitine on mitochondrial function in human embryos and embryo development. METHODS: To examine the effect of L-carnitine on mitochondrial function in morulae, 38 vitrified-thawed embryos at the 6-11-cell stage on day 3 after ICSI were donated from 19 couples. Each couple donated two embryos. Two siblings from each couple were divided randomly into two groups and were cultured in medium with or without 1 mM L-carnitine. The oxygen consumption rates (OCRs) were measured at morula stage. The development of 1029 zygotes cultured in medium with or without L-carnitine was prospectively analyzed. RESULTS: Addition of L-carnitine to the culture medium significantly increased the OCRs of morulae and improved the morphologically-good blastocyst formation rate per zygote compared with sibling embryos. Twenty healthy babies were born from embryos cultured in L-carnitine-supplemented medium after single embryo transfers. CONCLUSION(S): L-carnitine is a promising culture medium supplement that might be able to counteract the decreased mitochondrial function in human morula stage embryos.

Live birth outcomes of vitrified embryos generated under growth hormone stimulation are improved for women categorized as poor-prognosis / 대한생식의학회지

OBJECTIVE: To determine the clinical pregnancy (CP) and live birth (LB) rates arising from frozen embryo transfers (FETs) that had been generated under the influence of in vitro fertilization (IVF) adjuvants given to women categorized as poor-prognosis.METHODS: A registered, single-center, retrospective study. A total of 1,119 patients with first FETs cycle include 310 patients with poor prognosis (109 treated with growth hormone [GH], (+)GH group vs. 201 treated with dehydroepiandrosterone, (–)GH group) and 809 patients with good prognosis (as control, (–)Adj (Good) group).RESULTS: The poor-prognosis women were significantly older, with a lower ovarian reserve than the (–)Adj (Good) group, and demonstrated lower chances of CP (p<0.005) and LB (p<0.005). After adjusting for confounders, the chances of both CP and LB in the (+)GH group were not significantly different from those in the (–)Adj (Good) group, indicating that the poor-prognosis patients given GH had similar outcomes to those with a good prognosis. Furthermore, the likelihood of LB was significantly higher for poor-prognosis women given GH than for those who did not receive GH (p<0.028). This was further confirmed in age-matched analyses.CONCLUSION: The embryos cryopreserved from fresh IVF cycles in which adjuvant GH had been administered to women classified as poor-prognosis showed a significant 2.7-fold higher LB rate in subsequent FET cycles than a matched poor-prognosis group. The women with a poor prognosis who were treated with GH had LB outcomes equivalent to those with a good prognosis. We therefore postulate that GH improves some aspect of oocyte quality that confers improved competency for implantation.

Association between ovarian stimulators with or without intrauterine insemination, and assisted reproductive technologies on multiple births.

OBJECTIVE: We sought to quantify the risk of multiple births associated with the use of different modalities of medically assisted reproduction. STUDY DESIGN: We conducted a case-control study using a birth cohort from 2006 through 2009. This cohort was built with the linkage of data obtained by a self-administered questionnaire and medical, hospital, pharmaceutical, birth, and death databases in Quebec. Cases were pregnancies resulting in multiple live births (International Classification of Diseases, Ninth Revision/International Statistical Classification of Diseases, 10th Revision codes). Each case was matched, on maternal age and year of delivery, with 3 singleton pregnancies (controls) randomly selected among all Quebec singleton pregnancies. Data on the use of different fertility treatments were collected by a self-administered questionnaire. Multiple logistic regression models, adjusted for body mass index, number of previous live births, ethnicity, family income, place of residence, marital status, subfertility, reduction of embryos, diabetes, metformin treatment, folic acid supplementation, and lifestyle factors, were used to calculate the odds ratios (ORs) and confidence intervals (CIs). We evaluated the associations between each type of fertility treatment (ovarian stimulators used alone, intrauterine insemination [IUI] used with ovarian stimulation, and assisted reproductive technologies [ART]) and the risk of multiple births. RESULTS: A total of 1407 cases of multiple births and 3580 controls were analyzed. More than half of multiple births following medically assisted reproduction (53.6%) occurred among women having used ovarian stimulation with or without IUI. The use of ovarian stimulators alone and IUI with ovarian stimulation increase the risk of multiple births (adjusted OR, 4.5; 95% CI, 3.2-6.4; and adjusted OR, 9.32; 95% CI, 5.60-15.50, respectively) compared to spontaneous conception. The use of invasive ART was associated with a greatly increased risk of multiple births. Among only the 465 women who used medically assisted reproduction for conception, the use of IUI with ovarian stimulation was associated with an increased risk of multiple births (adjusted OR, 1.98; 95% CI, 1.12-3.49) when compared to ovarian stimulators used alone. Invasive ART were associated with an increased risk of multiple births (adjusted OR, 6.81; 95% CI, 3.72-12.49) when compared to ovarian stimulators used alone. CONCLUSION: Although the risk of multiple births associated with invasive ART can be decreased by elective implementing of single embryo transfer, special attention should be paid to the greatly increased risk associated with ovarian stimulation used alone or with IUI.

Evaluation of an effective multifaceted implementation strategy for elective single-embryo transfer after in vitro fertilization.

STUDY QUESTION: What is the relationship between the rate of elective single-embryo transfer (eSET) and couples' exposure to different elements of a multifaceted implementation strategy? SUMMARY ANSWER: Additional elements in a multifaceted implementation strategy do not result in an increased eSET rate. WHAT IS KNOWN ALREADY: A multifaceted eSET implementation strategy with four different elements is effective in increasing the eSET rate by 11%. It is unclear whether every strategy element contributes equally to the strategy's effectiveness. STUDY DESIGN AND SIZE: An observational study was performed among 222 subfertile couples included in a previously performed randomized controlled trial. PARTICIPANTS, SETTINGS AND METHODS: Of the 222 subfertile couples included, 109 couples received the implementation strategy and 113 couples received standard IVF care. A multivariate regression analysis assessed the effectiveness of four different strategy elements on the decision about the number embryos to be transferred. Questionnaires evaluated the experiences of couples with the different elements. MAIN RESULTS AND ROLE OF CHANCE: Of the couples who received the implementation strategy, almost 50% (52/109) were exposed to all the four elements of the strategy. The remaining 57 couples who received two or three elements of the strategy could be divided into two further classes of exposure. Our analysis demonstrated that additional elements do not result in an increased eSET rate. In addition to the physician's advice, couples rated a decision aid and a counselling session as more important for their decision to transfer one or two embryos, compared with a phone call and a reimbursement offer (P < 0.001). LIMITATIONS AND REASONS FOR CAUTION: The differences in eSET rate between exposure groups failed to reach significance, probably because of the small numbers of couples in each exposure group. WIDER IMPLICATIONS OF THE FINDINGS: Adding more elements to an implementation strategy does not always result in an increased effectiveness, which is in concordance with recent literature. This in-depth evaluation of a multifaceted intervention strategy could therefore help to modify strategies, by making them more effective and less expensive.

Arginine nutrition and fetal brown adipose tissue development in diet-induced obese sheep.

The global incidence of human obesity has more than doubled over the past three decades. An ovine model of obesity was developed to determine effects of maternal obesity and arginine supplementation on maternal, placental, and fetal parameters of growth, health, and well being. One-hundred-twenty days prior to embryo transfer, ewes were fed either ad libitum (n = 10) to induce obesity or 100% National Research Council-recommended nutrient requirements (n = 10) as controls. Embryos from superovulated ewes with normal body condition were transferred to the uterus of control-fed and obese ewes on day 5.5 post-estrus to generate genetically similar singleton pregnancies. Beginning on day 100 of gestation, obese ewes received intravenous administration of saline or L-arginine-HCl three times daily (81 mg arginine/kg body weight/day) to day 125, whereas control-fed ewes received saline. Fetal growth was assessed at necropsy on day 125. Maternal obesity increased (1) percentages of maternal and fetal carcass lipids and (2) concentrations of leptin, insulin, glucose, glutamate, leucine, lysine and threonine in maternal plasma while reducing (1) concentrations of progesterone, glycine and serine in maternal plasma and (2) amniotic and allantoic fluid volumes. Administration of L-arginine to obese ewes increased arginine and ornithine concentrations in maternal and fetal plasma, amniotic fluid volume, protein content in maternal carcass, and fetal brown adipose tissue (+60%), while reducing maternal lipid content and circulating leptin levels. Fetal or placental weight did not differ among treatments. Results indicate that arginine treatment beneficially reduces maternal adiposity and enhances fetal brown adipose tissue development in obese ewes.

Automated zebrafish chorion removal and single embryo placement: optimizing throughput of zebrafish developmental toxicity screens.

The potential of the developing zebrafish model for toxicology and drug discovery is limited by inefficient approaches to manipulating and chemically exposing zebrafish embryos-namely, manual placement of embryos into 96- or 384-well plates and exposure of embryos while still in the chorion, a barrier of poorly characterized permeability enclosing the developing embryo. We report the automated dechorionation of 1600 embryos at once at 4 h postfertilization (hpf) and placement of the dechorionated embryos into 96-well plates for exposure by 6 hpf. The process removed ≥95% of the embryos from their chorions with 2% embryo mortality by 24 hpf, and 2% of the embryos malformed at 120 hpf. The robotic embryo placement allocated 6-hpf embryos to 94.7% ± 4.2% of the wells in multiple 96-well trials. The rate of embryo mortality was 2.8% (43 of 1536) from robotic handling, the rate of missed wells was 1.2% (18 of 1536), and the frequency of multipicks was <0.1%. Embryo malformations observed at 24 hpf occurred nearly twice as frequently from robotic handling (16 of 864; 1.9%) as from manual pipetting (9 of 864; 1%). There was no statistical difference between the success of performing the embryo placement robotically or manually.