Classification of Bovine Cumulus-Oocyte Complexes with Convolutional Neural Networks

Determining oocyte quality is crucial for successful fertilization and embryonic development, there a serious correlation between live birth rates quality. Parameters such as the regular/irregular formation of cumulus cell layer around oocyte, number layers homogeneity appearance ooplasm are used to determine oocytes be in vitro (IVF) intracytoplasmic sperm injection (ICSI) methods.
 In this study, classification processes have been carried out using convolutional neural networks (CNN), deep learning method, on images cumulus-oocyte complex selected based theoretical knowledge professional experience embryologists. A network with depth 4 used. each level, one convolution, ReLU max-pooling included. The designed architecture trained Adam optimization algorithm. complexes (n=400) study were obtained by aspiration method from ovaries bovine slaughtered at slaughterhouse.
 CNN-based model developed showed promising results classifying three-class image data terms classification. achieved high accuracy, precision, sensitivity values test dataset. Continuous research can further improve its performance benefit field assessment.