Abstract: 【Objective】 The purpose of this study was to establish a non-destructive,non-invasive and rapid method for determining the sex of embryos.【Method】 207 porcine intracytoplasmic sperm injection (ICSI) embryos and their culture medium were used as the research objects.After the sex of a single embryo was identified by nest PCR amplification,it was randomly divided into training set and test set.At the same time,the Raman spectrum of a single ICSI embryo culture medium was obtained by Raman spectroscopy.After the original spectrum was processed,the support vector machine (SVM) algorithm was used to construct the classification model.Firstly,the training samples were used for training and modeling,and then the gender of the test samples was predicted.Combined with the gender identification results of PCR,the classification accuracy was calculated.【Result】 The sex of 207 embryos was indentified by nest PCR, 71 male and 128 female was successfully identified,and 8 samples failed to be amplified.The characteristic peak intensity of porcine ICSI male embryo culture medium at 1 082 and 1 360 cm^-1 Raman shifts was significantly higher than that of female embryo’s.The classification accuracy of the constructed embryo sex identification model was 81.5%,and the classification accuracy of female and male embryos were 81.3% and 81.8% respectively.【Conclusion】 The new and non-destructive method of embryo sex identification based on the Raman spectroscopy and SVM algorithm was constructed,and its classification accuracy was 81.5%.

Key words: Raman spectroscopy; support vector machine (SVM); embryo sex identification