time-lapse 培养和囊胚移植后胎心妊娠的预测工具---深度学习模型
发布时间:2020-12-21 19:09:08


近几年来,胚胎时差培养技术(time-lapse incubation)在辅助生殖治疗领域取得了重大突破,改变了常规胚胎培养模式。已经有大量研究文献证明了EmbryoScope时差培养箱在临床上的安全性和有效性,但是目前需要利用辅助注释(Guided Annotation)工具自动地提示所选的变量,并由胚胎学家进行确认,使用KIDScore决策支持工具为胚胎评估预测胎心妊娠的可能性,选择出着床潜能最高的胚胎进行移植或者冷冻。


我们创建了一个名为IVY的深度学习模型,该模型是一种客观且完全自动化的系统,可以从原始time-lapse视频中直接预测胎心妊娠的可能性,无需任何人工形态动力学注释或囊胚形态学评估。实验设计:回顾性分析了20141月至201812月期间,8个来自四个不同国家(地区)的IVF中心, 总计10 638个胚胎的实时成像和临床结果,使用具有已知胎心FH妊娠结果的实时成像视频对深度学习模型进行了模拟训练,在给定延时视频的情况下,执行预测FH怀孕可能性的二进制分类任务。深度学习模型能够根据实时成像视频的5层分层交叉验证中的AUC0.93 [95CI 0.92-0.94]来预测胎心FH妊娠。在八个不同的实验室进行的验证测试表明AUC具有可重现性,AUC范围为0.950.90。深度学习模型对胚胎植入具有很高的预测价值,可能会提高通过实时成像系统胚胎选择的方法的有效性, 可以改善单次胚胎移植的结果。 深度学习模型也可能被证明可为随后低温保存的胚胎的转移提供最佳顺序。


STUDY QUESTION: Can a deep learning model predict the probability of pregnancy with fetal heart (FH) from time-lapse videos?

SUMMARY ANSWER: We created a deep learning model named IVY, which was an objective and fully automated system that predicts the probability of FH pregnancy directly from raw time-lapse videos without the need for any manual morphokinetic annotation or blastocyst morphology assessment.

WHAT IS KNOWN ALREADY: The contribution of time-lapse imaging in effective embryo selection is promising. Existing algorithms for the analysis of time-lapse imaging are based on morphology and morphokinetic parameters that require subjective human annotation and thus have intrinsic inter-reader and intra-reader variability. Deep learning offers promise for the automation and standardization of embryo selection.

STUDY DESIGN, SIZE, DURATION: A retrospective analysis of time-lapse videos and clinical outcomes of 10 638 embryos from eight different IVF clinics, across four different countries, between January 2014 and December 2018.

PARTICIPANTS/MATERIALS, SETTING, METHODS: The deep learning model was trained using time-lapse videos with known FH pregnancy outcome to perform a binary classification task of predicting the probability of pregnancy with FH given time-lapse video sequence.The predictive power of the model was measured using the average area under the curve (AUC) of the receiver operating characteristic curve over 5-fold stratified cross-validation.

MAIN RESULTS AND THE ROLE OF CHANCE: The deep learning model was able to predict FH pregnancy from time-lapse videos with an AUC of 0.93 [95% CI 0.92–0.94] in 5-fold stratified cross-validation. A hold-out validation test across eight laboratories showed that the AUC was reproducible, ranging from 0.95 to 0.90 across different laboratories with different culture and laboratory processes.

LIMITATIONS, REASONS FOR CAUTION: This study is a retrospective analysis demonstrating that the deep learning model has a high level of predictability of the likelihood that an embryo will implant. The clinical impacts of these findings are still uncertain. Further studies, including prospective randomized controlled trials, are required to evaluate the clinical significance of this deep learning model. The time-lapse videos collected for training and validation are Day 5 embryos; hence, additional adjustment would need to be made for the model to be used in the context of Day 3 transfer.

WIDER IMPLICATIONS OF THE FINDINGS: The high predictive value for embryo implantation obtained by the deep learning model may improve the effectiveness of previous approaches used for time-lapse imaging in embryo selection. This may improve the prioritization of the most viable embryo for a single embryo transfer. The deep learning model may also prove to be useful in providing the optimal order for subsequent transfers of cryopreserved embryos.

STUDY FUNDING/COMPETING INTEREST(S): D.T. is the co-owner of Harrison AI that has patented this methodology in association with Virtus Health. P.I. is a shareholder in Virtus Health. S.C., P.I. and D.G. are all either employees or contracted with Virtus Health. D.G. has received grant support from Vitrolife, the manufacturer of the Embryoscope time-lapse imaging used in this study. The equipment and time for this study have been jointly provided by Harrison AI and Virtus Health.