Embryo selection in developing countries- inner cell mass to blastocyst dimension ratio as a predictor of pregnancy outcome

Abstract Study question Is the AI-based Life Whisperer™ (LW) tool, suitable to evaluate blastocysts quality and predict clinical pregnancy (CP) in couples undergoing ICSI cycles? Summary answer LW blastocyst score is comparable to the scores of other classification methods. This AI model showed high sensitivity and a comparable specificity for CP. What is known already The morphology grading is the most widely used method for the selection and classification of the embryos in clinical practice.However,this evaluation entails intervariability and intravariability decision among the embryologists.Recently, research has been focused on new embryo selection systems based on computer-assisted evaluation such as time-lapse with complex algorithms that allow the recognition of objective parameters of the embryo morphology.The implementation of these technologies requires substantial investments that are not available for all clinics.LW is a new embryo selection method based on AI,where specific hardware is not needed,as it is based on single blastocyst images taken with a routine microscope. Study design, size, duration Between 2017–2020, a total of 513 Day–5 blastocysts, after ICSI, comming from egg donation treatment were included in this retrospective-multicentre study.Day–5 embryos were evaluated with 3 classification methods:Gardner’s blastocyst grade (GB), the computer derived-output Eeva (EV) and LW AI-supported system. The good quality blastocysts were first evaluated using the GB and EV scores and subsequently compared with the LW scores.The sensitivity and specificity of LW was assessed to validate this system as a clinical pregnancy predictor. Participants/materials, setting, methods A total of 513 Day–5 blastocysts, from 134 oocyte donation cycles, were evaluated first by GB score: expansion (1–6), inner cell mass and throphoectoderm (A-C).EV analyses the cell division timing P2 (2cells stage duration) and P3 (3cells stage duration) differentiating three categories:High,Medium and Low(VerMilyea et al.,2014).LW scores ranked 1–10 from a single Day–5 blastocyst HR Image performed on inverted microscope,with a threshold >5 for defining a viable blastocyst.T-test and ROC-curves were used for statistical analysis. Main results and the role of chance The average of LW score obtained from GB higher blastocyst expansion score (≥4) was 7.48±0.09, while the average of LW score obtained from GB lower blastocyst expansion score (<4) was 4.69±0.3 (P < 0.001). The average of LW score yielded from GB good morphology of Inner Cell Mass and trophoectoderm (AA,AB,BA) was 7.98±0.1 while the average of LW score obtained from GB lower quality blastocyst score (BB,BC,CB,CA,AC) was 6.36±0.156 (P < 0.001).The average of LW score resulted from EV High blastocysts was 7.42±0.17, while the average of this obtained from EV low score was 6.43±0.3 (P = 0.009).A correlation between EV and LW score could be assesed, except for the blastocyst that are considered Medium score from EV. Therefore, a strong correlation between GB and LW system, as well GB+EV and LW, was found and an equivalent usability of the LW tool could be confirmed. The analyse of LW score for transferred embryos (N = 156), using ROC curve, showed a high sensitivity (0,928) but a low specificity (0,154) with a threshold of 5. Regarding our data, ROC curve shows that a threshold of 8,46 could enhance the prediction of CPR because in this point the specifity value is higher than 0.5. Limitations, reasons for caution The LW score validation compared to GB and EV methodology was carried out on a small number of embryos.Additionally,not all embryos had been transferred at the time of the analysis.Thus to enhance the accuracy of these data and the specificity of the clinical prediction, a higher sample size is needed. Wider implications of the findings: Blastocyst selection looks equivalent between all systems,but the LW tool is more objective and faster, saving time and costs significantly, without needing substantial hardware investments. Additionally,the LW-system shows almost the highest sensibility and may also improve the specificity by self-learning feeding the AI-system, thus tailoring predictions to each laboratory unique environment. Trial registration number NA

Download Full-text

Overall Blastocyst Quality, Trophectoderm Grade, and Inner Cell Mass Grade Predict Pregnancy Outcome in Euploid Blastocyst Transfer Cycles

Chinese Medical Journal ◽

10.4103/0366-6999.232808 ◽

2018 ◽

Vol 131 (11) ◽

pp. 1261-1267 ◽

Cited By ~ 13

Author(s):

Yan-Yu Zhao ◽

Yang Yu ◽

Xiao-Wei Zhang

Keyword(s):

Pregnancy Outcome ◽

Inner Cell Mass ◽

Cell Mass ◽

Blastocyst Transfer ◽

Blastocyst Quality

Download Full-text

P–163 The presence of trophectodermal vesicles on expanded ICSI-derived blastocysts is a good predictor of implantation

Human Reproduction ◽

10.1093/humrep/deab130.162 ◽

2021 ◽

Vol 36 (Supplement_1) ◽

Author(s):

O Pravdyuk ◽

M Gryshchenko ◽

L Shatalova ◽

K Borodai

Keyword(s):

Zona Pellucida ◽

Inner Cell Mass ◽

Cell Mass ◽

Implantation Rate ◽

Embryo Selection ◽

Assisted Hatching ◽

Ratio Test ◽

Group 3 ◽

Group 2 ◽

Group 1

Abstract Study question Is the implantation rate (IR) higher in blastocysts with trophectodermal vesicles (TVs) compared to blastocysts without TVs or euploid blastocysts with unknown spontaneous hatching status? Summary answer The blastocysts with TVs demonstrate significantly higher IR in comparison to blastocysts without TVs or euploid blastocysts with unknown spontaneous hatching status. What is known already After ICSI spontaneous hatching mainly occurs by trophectoderm cell herniation via a small slit in the zona pellucida. At the beginning of this process, TVs are formed on the outside of the zona pellucida. It was previously shown that the clinical pregnancy rate was similar after the transfer of expanded blastocysts and expanded blastocysts with TVs. But another study showed that transfer of blastocysts of more advanced hatching stages yields better pregnancy rates than expanded blastocyst transfer. It remains unclear whether there is an association between the presence of TVs and IR in the transfers of single vitrified blastocysts. Study design, size, duration This retrospective cohort study was conducted from October 2018 to November 2019 and included 477 transfers of a single vitrified blastocyst. Cases were divided into 3 groups. Group 1 included transfers of blastocysts without TVs and with assisted hatching (AH). Group 2 contained the transfers of blastocysts at TVs stage of spontaneous hatching and without AH. Group 3 consisted of transfers of the euploid blastocysts with AH performed and unknown spontaneous hatching status. Participants/materials, setting, methods The age of women was between 21 and 39. Embryo transfers following oocyte donation programs were excluded from the study. This study included only transfers of the ICSI-derived fully expanded blastocysts with top-graded inner cell mass and trophectoderm. AH was performed using laser Saturn 3. The primary outcome was the implantation rate. Statistical analysis was performed using Pearson’s chi-square test and likelihood ratio test. Preimplantation genetic testing for aneuploidy (PGT-A) was performed by next-generation sequencing. Main results and the role of chance The number of cases in groups 1,2 and 3 was 133, 49, and 295, respectively. The average age in the groups was about 32.5 and did not differ between groups. The implantation rate in group 3 with PGT-A was 60% (177 out of 295), which was insignificantly higher compared to group 1 - 55% (73 out of 133) (p = 0.34). In group 2, the implantation rate was 76% (38 out of 49), which exceeded significantly the outcomes in groups 1 and 2 (p = 0.016). Thus the transfer of expanded blastocyst with TVs gives higher IR in comparison to expanded blastocyst. Therefore TVs could be utilized as a morphological marker for embryo selection. Furthermore, according to obtained results the presence of TVs on nontested blastocysts predicts implantation better than euploidy does in blastocysts with unknown spontaneous hatching status. Limitations, reasons for caution This is a retrospective nonrandomized study with its inherited limitations. Wider implications of the findings: Based on the results of the study embryo selection practice could be optimized. To maximize the outcomes of PGT-A programs embryo culture and biopsy workflow could be modified to allow collecting data on spontaneous hatching and TVs presence before performing the biopsy. Trial registration number Not applicable

Download Full-text

P–260 Towards better explainable deep learning models for embryo selection in ART

Human Reproduction ◽

10.1093/humrep/deab130.259 ◽

2021 ◽

Vol 36 (Supplement_1) ◽

Author(s):

A Sharma ◽

T Haugen ◽

H Hammer ◽

M Rieleger ◽

M Stensen

Keyword(s):

Deep Learning ◽

Network Architecture ◽

Inner Cell Mass ◽

Visual Representation ◽

Cell Mass ◽

Quantitative Measure ◽

Embryo Selection ◽

Training Data ◽

Embryo Viability ◽

Filter Size

Abstract Study question Can heatmaps generated by occlusion explain the patterns learned by deep learning (DL) models classifying the embryo viability in ART? Summary answer Occlusion experiments generate heatmaps that reveal which regions in frames of time-lapse video (TLV) are more discriminative for classification and prediction by the DL models. What is known already DL has widely been explored in ART for embryo selection. Depending upon input (video or image), different DL models classifying embryo viability are developed. However, whether the prediction is based on actual input features or random guessing is unknown. The embryo selection in ART is subjective. If the intention is using DL models’ prediction to transfer, freeze or discard the embryo, explanations of how they interpret embryonic development features brings transparency and trust. In other areas, heatmaps are used for explaining DL predictions. The heatmaps can be a tool to understand patterns learned by DL models for embryo selection. Study design, size, duration We trained two separate DL models for predicting the presence of fetal heartbeat for the transferred embryos. We further used occlusion generated heatmaps to explain the predictions. For training, retrospective data was used. The input dataset consisted of 136 TLVs and corresponding patient data for 132 participants (128: single embryo transfers and 8: double embryo transfer) from both IVF and ICSI treatment. Each video was assessed by an embryologist. Participants/materials, setting, methods DL models (A as ResNet–18, B as VGG16) are trained for predicting the presence of fetal heartbeat on a single frame extracted from TLV after day three or later. Model A has a better recall (0.7) compared to B (0.5). Heatmaps explain the reason behind models’ recall rate by visually representing patterns learned by them. Using occlusion filter size 30*30 with stride 14 and size 50*50 with stride 25, we generate heatmaps for both models. Main results and the role of chance The heatmaps generated using occlusion can represent visually the patterns discovered by the DL models when predicting the presence of a fetal heartbeat. Using occlusion filter size 30*30 with stride 14, we verified that Model B has lower recall because the heatmaps show that the model finds redundant features present outside the embryo region in many input frames. It could be interpreted that either the model has not learned relevant patterns or is more robust to noise. This representation of DL models equips us in better decision-making, whether to consider or discard the prediction or rather train the model further, preprocess training data or change network architecture. The heatmaps revealed that for frames where significant patterns learned by the models are within the embryo region, more weight was given to specific features like the inner cell mass, trophectoderm and some parts within the zona pellucida. Moreover, the heat maps generated using occlusion are independent of the underlying model’s architecture as the same experiment settings were used for both models. For occlusion filter size 50*50 with stride 25, the expanse of input regions (in or outside the embryo) considered relevant could be visualized for both models A and B. Limitations, reasons for caution Heatmaps generated by occluding input regions give a visual representation of features in individual frames not directly on videos. Explaining DL models by heatmaps besides occlusion, other techniques (Grad-Cam) exist but were not evaluated. Furthermore, there is no quantitative measure for evaluating whether heatmaps are a good explanation or not. Wider implications of the findings: The heatmaps make the patterns discovered by DL models visually recognized and bring forth the prominent portions of embryo regions. This will again improve understanding and trust in DL models’ predictions. Visual representation of DL models using heatmaps enables interpreting a prediction, performing model analysis and determining scope for improvement. Trial registration number Not applicable

Download Full-text

Use of F9 teratocarcinoma STEM cells to study cell-matrix interactions in the early mouse embryo

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s0424820100123416 ◽

1992 ◽

Vol 50 (1) ◽

pp. 602-603

Author(s):

Marc Lenburg ◽

Rulang Jiang ◽

Lengya Cheng ◽

Laura Grabel

Keyword(s):

Mouse Embryo ◽

Inner Cell Mass ◽

Cell Mass ◽

Cell Types ◽

Embryoid Bodies ◽

F9 Cells ◽

Cell Matrix ◽

Matrix Interactions ◽

Cell Matrix Interactions ◽

F9 Teratocarcinoma

We are interested in defining the cell-cell and cell-matrix interactions that help direct the differentiation of extraembryonic endoderm in the peri-implantation mouse embryo. At the blastocyst stage the mouse embryo consists of an outer layer of trophectoderm surrounding the fluid-filled blastocoel cavity and an eccentrically located inner cell mass. On the free surface of the inner cell mass, facing the blastocoel cavity, a layer of primitive endoderm forms. Primitive endoderm then generates two distinct cell types; parietal endoderm (PE) which migrates along the inner surface of the trophectoderm and secretes large amounts of basement membrane components as well as tissue-type plasminogen activator (tPA), and visceral endoderm (VE), a columnar epithelial layer characterized by tight junctions, microvilli, and the synthesis and secretion of α-fetoprotein. As these events occur after implantation, we have turned to the F9 teratocarcinoma system as an in vitro model for examining the differentiation of these cell types. When F9 cells are treated in monolayer with retinoic acid plus cyclic-AMP, they differentiate into PE. In contrast, when F9 cells are treated in suspension with retinoic acid, they form embryoid bodies (EBs) which consist of an outer layer of VE and an inner core of undifferentiated stem cells. In addition, we have established that when VE containing embryoid bodies are plated on a fibronectin coated substrate, PE migrates onto the matrix and this interaction is inhibited by RGDS as well as antibodies directed against the β1 integrin subunit. This transition is accompanied by a significant increase in the level of tPA in the PE cells. Thus, the outgrowth system provides a spatially appropriate model for studying the differentiation and migration of PE from a VE precursor.

Download Full-text