拉曼光谱在胚胎质量评估中的应用价值  

Application value of Raman spectroscopy on embryo quality assessment

作　　者：黄静 姜宏[1] 刘颖[1] 刘迎春[1] 倪丰 HUANG Jing;JIANG Hong;LIU Ying;LIU Ying-chun;NI Feng(Reproductive Medicine Center,the 901 th Hospital of the Joint Logistics Support Force of PLA,Hefei 230031)

机构地区：[1]中国人民解放军联勤保障部队第九○一医院生殖中心,合肥230031

出　　处：《生殖医学杂志》2021年第8期1076-1081,共6页Journal of Reproductive Medicine

基　　金：安徽省科技攻关计划项目(1604a0802095)。

摘　　要：目的探讨拉曼光谱在胚胎质量评估中的应用价值。方法收集2018年5月至2019年1月在我院生殖中心行卵胞浆内单精子注射(ICSI)治疗的16例患者D3胚胎培养液共122份,采用传统的形态学评分方法对D3胚胎进行评估,将其分为优质胚胎组(66份)与非优质胚胎组(56份),利用拉曼光谱法结合传统的化学计量学方法中主成分分析(PCA)分别对两组培养液进行分析,结合机器学习算法[卷积神经网络模型(CNN)]对所获得的拉曼数据进行模型构建及结果预测。结果优质胚胎组和非优质胚胎组的拉曼光谱在第一维主成分(PC1)有显著性差异(P<0.05),在第二维主成分(PC2)和第三维主成分(PC3)均无显著性差异(P>0.05)。CNN算法构建模型及预测结果显示,拉曼光谱预测优质胚胎的特异性为71.21%,敏感性为73.21%,准确性为72.13%。结论优质胚胎与非优质胚胎的拉曼光谱明显不同,采用拉曼光谱结合PCA分析及CNN算法,可以区分优质胚胎与非优质胚胎,对胚胎质量评估有一定的参考价值。Objective:To investigate the application value of Raman spectroscopy on embryo quality assessment.Methods:A total of 122 samples of culture media of Day 3 embryos from 16 women who underwent ICSI in the Reproductive Medicine Center of our hospital were collected.The conventional morphological scoring method was used to evaluate the Day 3 embryos,and the embryos were divided into the good-quality embryo group(66 samples)and the poor-quality embryo group(56 samples)according to the results of morphological evaluation.Then the sample of culture media of the two groups were analyzed by using Raman spectroscopy combined with the traditional chemometrics method(Principal Component Analysis,PCA).The acquired Raman data were modeled and predicted by the advanced machine learning algorithm[Convolutional Neural Network(CNN)].Results:There was significantly difference in principal component in the first dimension(PC1)of Raman spectroscopy between good-quality embryo group and poor-quality embryo group(P<0.05),while the second-dimensional principal component(PC2)and the third-dimensional principal component(PC3)was not significantly different(P>0.05).The results of CNN algorithm showed that the specificity of Raman spectroscopy predicting good-quality embryo was 71.21%,the sensitivity was 73.21%,and the accuracy was 72.13%.Conclusions:Raman spectroscopy combined with PCA analysis and CNN algorithm can distinguish good-quality embryos from poor-quality embryos,which has a certain reference value for embryo quality evaluation.

关 键 词：辅助生殖技术 拉曼光谱 胚胎培养液 卵胞浆内单精子注射 胚胎质量 

分 类 号：R711.6[医药卫生—妇产科学]