40 Bovine embryo survival of cryopreservation detected through analysis of real-time video

Cryopreservation of both in vivo-derived (IVD) and in vitro-produced (IVP) embryos allow genetic storage, recipient management and transport of genetic material. However, cryopreservation can induce cryodamage or cryotoxicity. To maximize the benefits of embryo cryopreservation to the livestock industry, prevention, and detection of cryodamage is key. The objective of this study was to use graphic image processing (GIP) and machine learning (ML) techniques to evaluate embryos post-cryopreservation to generate predictive models to detect embryo survival of cryopreservation. One hundred twenty-seven bovine IVD embryos were frozen (slow freeze protocol) either in glycerol (n = 63) or ethylene glycol (EG; n = 64) during routine processing of embryos. Embryos were thawed (held in air 5 s, plunged in 29°C water for 15 s and unloaded). Embryos treated with glycerol were transferred into holding medium whereas embryos treated with EG were first transferred into ABT 1 Step Thaw Solution, then holding medium. After allowing 1 h for rehydration, embryos were graded according to International Embryo Transfer Society (IETS) standards and video recorded for 30 s with a smartphone mounted to a stereoscope. After one video recording per embryo, embryos were placed in single culture droplet and cultured for 48 h. Embryos were re-evaluated and photographed (single frame) at 24 h and 48 h. Videos were processed with GIP techniques to quantify frame-by-frame pixel change throughout the 30-s video duration. This allows embryo morphokinetics to be measured over time, which was presented as normalized activity (standard deviation/mean) pixel change. Comparisons between the level of morphokinetic activity present in embryos which advanced in stage vs embryos which failed to advance in stage were recorded. Data were analysed with a Student’s t-test and chi-squared in R. There was no statistical difference in developmental outcomes from each treatment group as 43/63 (68.25%) of embryos treated with glycerol advanced in developmental stage and 36/64 (56.25%) embryos frozen in ethylene glycol advanced in developmental stage (P > 0.05). The GIP processing showed embryos that advanced had more activity than embryos that stalled (P < 0.05). Additionally, embryos frozen in glycerol had higher morphokinetic activity, through results were not significant (P > 0.05). Interestingly, in both treatment groups, most of the embryos that failed to develop clustered 7–17% below the mean. Embryos that survived cryopreservation, evidenced by continued development, have higher morphokinetic activity levels likely due to increased cellular activity and metabolism. It was also observed that the embryos frozen in glycerol demonstrated higher activity levels, but as these embryos had a higher development rate, this observation could be attributed to the high proportions of developing, metabolically active embryos. In conclusion, the detection of this biological phenomena can allow embryologists to utilise objective metrics to evaluate embryo health and select which embryos are eligible for transfer and improve live birth outcomes of cryopreserved embryos. Future work aims to develop ML algorithms to detect embryo survival of cryopreservation for use by embryologists.