Robotic denudation of zygotes

Abstract

The vitrification technology is used in the embryo freezing process. To ensure intracellular water removal and avoid ice crystal formation, the cumulus cells surrounding zygotes need to be removed. As operators have varying skill and long-time fatigue, it is challenging to completely remove the cumulus cells surrounding zygotes and reduce zygote loss inside the micropipette. The objective of this work is to develop a robotic zygote denudation system for the vitrification procedure. The robotic system enables accurate segmentation of cumulus cells for estimating the mass of cumulus zygote complexes (CZCs) inside the micropipette via the Res-Unet neural network. A mathematical model was built to describe the dynamic motion of CZCs inside the micropipette, and a model-based optimal controller was developed to aspirate and deposit CZCs inside the micropipette. To denude zygote completely, the Res-Attention neural network was used to detect cumulus cells for predicting the quality of zygote denudation. In the mouse zygotes experiments, the yield rate is 98.2% ± 1.6% and the denudation efficacy is 96.9% ± 0.2% for the robotic system. Compared with manual denudation, the survival rate and development rate of embryos cultured from zygotes denuded by the robotic system were both higher in subsequent vitrification procedure.