8 New techniques for small ruminant embryo evaluation and selection in routine practice

Advanced embryo collection techniques such as laparoscopic ovum pickup combined with in vitro embryo production or transcervical uterine flushing in small ruminant species allows for the increased proliferation of genetics from superior females. Still, methods of embryo selection have largely remained static, which has limited the growth and efficiency of embryo transfer. Morphokinetic analysis has emerged as a noninvasive measurement to determine the quality of embryos. When combined with advanced computer vision technologies, this analysis can consistently provide objective data relevant to developmental potential. In other species, embryo morphokinetic activity has revealed information about embryo stress and metabolism and has linked embryos to pregnancy outcomes from short real-time videos. Thus, the objective of this study was to determine if advanced computer techniques could be applied to detect embryo morphokinetic activity from short real-time videos of caprine and ovine embryos collected in routine practice. A 30-s video was recorded of embryos using standard cell phone and microscope equipment. Object detection and background subtraction computer vision techniques were applied to videos to quantify action area pixel changes. Action areas (n = 172) were used to identify individual embryos and randomly generate control areas within each video that did not contain embryos. Action areas for both groups were subjected to the background subtraction process to generate computer identified pixel changes for quantification of activity analysis. The embryos were stage and quality graded by an experienced embryologist in accordance with IETS standards. Data analyzed using ANOVA with a P-value of 0.05 for significance. Results are shown as (mean ± standard deviation) of computer-identified activity levels derived from object detection and background subtraction for action areas of 30-s videos. Action areas that contained embryos appeared to have greater identified activity (0.592 ± 0.2) than the control action areas (0.302 ± 0.01; P < 0.05). Preliminary data provide evidence that computer vision technology could be used to identify activity differences of small ruminant embryos from video analysis captured on a smart phone connected to a microscope. It is hypothesized that computer-identified activity from the action areas with embryos is reflective of embryo cellular activity that is largely nonidentifiable to the human eye under the microscope, while the control areas may have some identified activity from additional matter that is moving in the collection dish at the time of video. These results provide a foundation for additional work to be done evaluating small ruminant embryo morphokinetics and important associated developmental and pregnancy outcomes.

Partial funding was provided by the National Science Foundation.