Improving Deep Learning-Based Algorithm for Ploidy Status Prediction Through Combined U-NET Blastocyst Segmentation and Sequential Time-Lapse Blastocysts Images.

Background: Several approaches have been proposed to optimize the construction of an artificial intelligence-based model for assessing ploidy status. These encompass the investigation of algorithms, refining image segmentation techniques, and discerning essential patterns throughout embryonic development. The purpose of the current study was to evaluate the effectiveness of using U-NET architecture for embryo segmentation and time-lapse embryo image sequence extraction, three and ten hr before biopsy to improve model accuracy for prediction of embryonic ploidy status. Methods: A total of 1.020 time-lapse videos of blastocysts with known ploidy status were used to construct a convolutional neural network (CNN)-based model for ploidy detection. Sequential images of each blastocyst were extracted from the time-lapse videos over a period of three and ten hr prior to the biopsy, generating 31.642 and 99.324 blastocyst images, respectively. U-NET architecture was applied for blastocyst image segmentation before its implementation in CNN-based model development. Results: The accuracy of ploidy prediction model without applying the U-NET segmented sequential embryo images was 0.59 and 0.63 over a period of three and ten hr before biopsy, respectively. Improved model accuracy of 0.61 and 0.66 was achieved, respectively with the implementation of U-NET architecture for embryo segmentation on the current model. Extracting blastocyst images over a 10 hr period yields higher accuracy compared to a three-hr extraction period prior to biopsy. Conclusion: Combined implementation of U-NET architecture for blastocyst image segmentation and the sequential compilation of ten hr of time-lapse blastocyst images could yield a CNN-based model with improved accuracy in predicting ploidy status.

The origin and possible mechanism of embryonic cell-free DNA release in spent embryo culture media: a review.

The presence of cell-free DNA in spent embryo culture media (SECM) has unveiled its possible utilization for embryonic ploidy determination, opening new frontiers for the development of a non-invasive pre-implantation genetic screening technique. While a growing number of studies have shown a high concordance between genetic screening using cell-free DNA (cfDNA) and trophectoderm (TE), the mechanism pertaining to the release of cfDNA in SECM is largely unknown. This review aims to evaluate research evidence on the origin and possible mechanisms for the liberations of embryonic DNA in SECM, including findings on the self-correction abilities of embryos which might contribute to the presence of cfDNA. Several databases including EMBASE, PUBMED, and SCOPUS were used to retrieve original articles, reviews, and opinion papers. The keywords used for the search were related to the origins and release mechanism of cfDNA. cfDNA in SECM originates from embryonic cells and, at some levels, non-embryonic cells such as maternal DNA and exogenous foreign DNA. The apoptotic pathway has been demonstrated to eliminate aneuploid cells in developing mosaic embryos which might culminate to the release of cfDNA in SECM. Nonetheless, there is a recognized need for exploring other pathways such as cross-talk molecules called extracellular vesicles (EVs) made of small, round bi-layer membranes. During in vitro development, embryos physiologically and actively expel EVs containing not only protein and microRNA but also embryonic DNA, hence, potentially releasing cfDNA of embryonic origin into SECM through EVs.

Genetic algorithm-assisted machine learning for clinical pregnancy prediction in in vitro fertilization.

BACKGROUND: A clinical pregnancy prediction model was developed by implementing machine learning technology that uses a combination of static images and medical data to calculate the outcome of an in vitro fertilization cycle. OBJECTIVE: To provide a system that can accurately and sufficiently assist with decision making that is critical to in vitro fertilization cycles, primarily embryo selection. STUDY DESIGN: Historical medical data, which consist of clinical information and a complete transferred embryo image dataset, of 697 patients who underwent unique in vitro fertilization were collected. Various techniques of machine learning were used, namely decision tree, random forest, and gradient boosting; each technique used the same data configuration for performance comparison and was subsequently optimized using genetic algorithm. RESULTS: A prediction model with a peak accuracy of approximately 65% was achieved. Significant differences in the performances of the 3 selected algorithms were apparent. Nonetheless, additional metric measurements, such as receiver operating characteristic, area under the receiver operating characteristic curve score, accuracy, and loss, suggested that the gradient boosting model performed the best in predicting clinical pregnancy. CONCLUSION: This study served as a stepping stone toward the application of in vitro fertilization prediction models that use machine learning techniques. However, additional validation steps are required to boost the model's performance for its implementation in the clinical setting.

A Homogeneous Ensemble of Robust Pre-defined Neural Network Enables Automated Annotation of Human Embryo Morphokinetics.

Background: The purpose of the current study was to reduce the risk of human bias in assessing embryos by automatically annotating embryonic development based on their morphological changes at specified time-points with convolutional neural network (CNN) and artificial intelligence (AI). Methods: Time-lapse videos of embryo development were manually annotated by the embryologist and extracted for use as a supervised dataset, where the data were split into 14 unique classifications based on morphological differences. A compilation of homogeneous pre-trained CNN models obtained via TensorFlow Hub was tested with various hyperparameters on a controlled environment using transfer learning to create a new model. Subsequently, the performances of the AI models in correctly annotating embryo morphologies within the 14 designated classifications were compared with a collection of AI models with different built-in configurations so as to derive a model with the highest accuracy. Results: Eventually, an AI model with a specific configuration and an accuracy score of 67.68% was obtained, capable of predicting the embryo developmental stages (t1, t2, t3, t4, t5, t6, t7, t8, t9+, tCompaction, tM, tSB, tB, tEB). Conclusion: Currently, the technology and research of artificial intelligence and machine learning in the medical field have significantly and continuingly progressed in an effort to develop computer-assisted technology which could potentially increase the efficiency and accuracy of medical personnel's performance. Nonetheless, building AI models with larger data is required to properly increase AI model reliability.

Role of three-dimensional Doppler ultrasonography and leukemia inhibitory factor from endometrial secretion in predicting endometrial receptivity in IVF treatment: a pilot study.

PURPOSE: This pilot study aimed to evaluate the potential synergistic role of three-dimensional power Doppler angiography ultrasound and the expression of Leukemia Inhibitory Factor (LIF) protein in predicting the endometrial receptivity of fresh In-Vitro Fertilization (IVF) cycles. MATERIALS AND METHODS: This prognostic cohort study involved 29 good prognosis women who underwent fresh IVF cycles with fresh blastocysts transfer. Serial measurements of sub-endometrial parameters including vascularity index (VI), flow index (FI), and vascularization flow index (VFI) were conducted consecutively via power Doppler angiography on the day of oocyte maturation trigger, oocyte retrieval, and blastocyst transfer. Aspiration of endometrial secretion was performed on the day of embryo transfer. RESULTS: The mean index of VI and VFI on the trigger and oocyte retrieval day and also LIF protein concentration at the window of implantation were significantly higher in clinically pregnant women than that of the non-pregnant women (p < 0.05). The area under the curve (AUC) of VI and VFI was shown to have a powerful predictive value to forecast receptive endometrium on either trigger day (0.788 and 0.813, respectively) or oocyte retrieval day (0.813 and 0.818). Likewise, LIF concentration on the day of embryo transfer was adequate to become a predictor for endometrial receptivity (AUC 0.874). A combination of the VI and VFI on the trigger day and LIF concentration at specific cut-off values (VI > 5.381, VFI > 1.483, LIF 703.5 pg/mL) produced an algorithm with high AUC (0.881) and high specificity (94.4%) for an adequate prediction of non-receptive endometrium. CONCLUSION: VI and VFI index assessed on maturation trigger day and the expression of LIF protein concentration at the window of implantation provided sufficient information to predict endometrial receptivity. A large randomized control trial is needed to validate these findings.

Review of computer vision application in in vitro fertilization: the application of deep learning-based computer vision technology in the world of IVF.

In vitro fertilization has been regarded as a forefront solution in treating infertility for over four decades, yet its effectiveness has remained relatively low. This could be attributed to the lack of advancements for the method of observing and selecting the most viable embryos for implantation. The conventional morphological assessment of embryos exhibits inevitable drawbacks which include time- and effort-consuming, and imminent risks of bias associated with subjective assessments performed by individual embryologists. A combination of these disadvantages, undeterred by the introduction of the time-lapse incubator technology, has been considered as a prominent contributor to the less preferable success rate of IVF cycles. Nonetheless, a recent surge of AI-based solutions for tasks automation in IVF has been observed. An AI-powered assistant could improve the efficiency of performing certain tasks in addition to offering accurate algorithms that behave as baselines to minimize the subjectivity of the decision-making process. Through a comprehensive review, we have discovered multiple approaches of implementing deep learning technology, each with varying degrees of success, for constructing the automated systems in IVF which could evaluate and even annotate the developmental stages of an embryo.

Morphokinetics of embryos after IMSI versus ICSI in couples with sub-optimal sperm quality: A time-lapse study.

An investigation was conducted to determine the influence of two sperm selection modalities, IMSI and ICSI, on the morphokinetics, dynamic development and ploidy status of embryos derived from males with sub-optimal sperm profiles during IVF program. A total of 209 PGTA-tested top-quality blastocysts (IMSI = 129, ICSI = 80) from 84 couples (IMSI = 51, ICSI = 33) were assessed retrospectively. This study found that both IMSI and ICSI yielded comparable embryo morphokinetics, except for the T7, TEB and CC3 parameters (p < 0.05). A significant lower incidence of multinucleation was observed in the IMSI group when compared to the ICSI group (48.8% vs. 71.3%, p = 0.002), while other parameters of embryo development such as direct cleavage, distorted cytoplasmic movement, reverse cleavage and vacuole(s) appearance did not differ (p > 0.05). No differences were noticed in the proportion of generating chromosomally euploid embryos (44.2% vs. 51.3%, p = 0.394, respectively, for IMSI and ICSI). The implementation of IMSI or ICSI in couples with sub-optimal sperm profiles resulted in embryos with comparatively similar morphokinetics. Furthermore, the incidence of multinucleation at the two- to four-cell stage was lower following the practice of IMSI, although the method did not improve the proportion of gaining euploid embryos.

Embryo Banking with Mild Ovarian Stimulation for IVF: An Alternative Strategy for Poor Prognosis Patients.

BACKGROUND: Management of Poor Ovarian Reserve (POR) in in vitro fertilization remains a difficult challenge. The purpose of this retrospective cohort study was to compare the effectiveness of embryo banking strategy over a cohort of several mild stimulation cycles (Embryo Banking Strategy for Poor Prognosis/Embargo) to conventional full-dose antagonist protocol for IVF. METHODS: Subjects identified as having poor ovarian response (POR) based on the Bologna criteria were recruited. In total, there were 113 subjects included in the analysis. Fifty-three subjects underwent embryo banking procedure (Embargo) protocol, and sixty subjects underwent the conventional full-dose antagonist protocol for IVF. The Chi-square test was used to compare the clinical pregnancy rate, miscarriage rate as well as live birth rate, while the Mann-Whitney U test was utilized to analyze the cost per clinical pregnancy between the two groups. A p<0.05 was considered statistically significant. RESULTS: The two studied groups showed similar outcomes regarding clinical pregnancy rate, miscarriage rate, as well as live birth rate (p=0.966, p=0.310, and p= 0.469, respectively). Cost analysis of subjects who underwent mild ovarian stimulation followed by Embargo revealed the high cost of the protocol compared to conventional full-dose antagonist protocol ($10.507±6.181 vs. $9.533±2.530, p=0.002). CONCLUSION: The clinical outcomes of both protocols were comparable. Embargo procedure was not efficient in improving the overall clinical outcomes in patients who were expected poor ovarian responders as the protocol costed more comparing with conventional full-dose antagonist protocol. A larger prospective randomized control trial is needed to evaluate this finding.