From Pixels to Pregnancies: AI-Driven Oocyte Grading for Scalable Livestock Breeding

Parkash Singh, Grace Koppelman, Tanya Berger-Wolf, John Patrick Fulton

Submission Track: Short papers presenting ongoing research or work submitted to other venues (up to 5 pages, excluding references)

Keywords: Assisted reproductive technologies (ART), in vitro fertilization (IVF), embryo development, blastocyst yield, cumulus–oocyte complex, embryo development, livestock breeding, International Embryo Transfer Society (IETS), Oocyte grading, livestock reproduction

Abstract: Sustainable livestock breeding is essential to meeting the food demands of a growing global population. Assisted reproductive technologies (ART), such as in vitro fertilization(IVF), are increasingly used to enhance reproductive efficiency. A key determinant of IVF success is the quality of the oocyte, which directly affects fertilization, embryo development, and blastocyst yield. However, oocyte grading today remains a subjective and inconsistent process, creating variability that affects the entire in vitro embryo production pipeline. We introduce a deep learning framework for automated oocyte grading, built on the first dataset of its kind: 1,140 bovine cumulus–oocyte complex (COC) images labeled according to the International Embryo Transfer Society (IETS) scale. Our models achieve up to 65\% accuracy across four grades, improving to over 80\% when grouped into industry relevant quality categories. By aligning with IETS guidelines, we establish the first benchmark for standardized oocyte grading in livestock IVF. This work provides a strong foundation for AI-based assisted livestock breeding, offering consistency, reduced human variability, and increased throughput in livestock breeding.

Submission Number: 67