P-067 The role of artificial intelligence (AI) in selecting sperm for intracytoplasmic sperm injection (ICSI): a pilot study
C Pastor Leary, M Hörmann-Kröpfl, M Schenk, G WeissAbstract
Study question
Can the use of AI for sperm selection during ICSI enhance embryological and clinical outcomes?
Summary answer
Oocytes inseminated with sperm selected through AI software yielded comparable embryological outcomes to conventional methods, with a significantly improved embryo utilization rate.
What is known already
Since its introduction in the 1990s, ICSI has become the cornerstone of assisted reproductive technology (ART). Numerous refinements, including IMSI and PICSI, have aimed to optimize outcomes. Recently, AI has emerged in the field of assisted reproductive techniques to enhance decision-making processes. SiD software employs an AI-based algorithm to assist sperm selection during ICSI. Although initial studies suggest its potential to improve outcomes, data on fertilization rates, embryo development, and clinical efficacy remain limited and warrant further investigation.
Study design, size, duration
This prospective, monocentric study was conducted at Kinderwunsch Institut Schenk GmbH (Austria) from January to May 2024. A total of 29 couples undergoing ICSI were included, sibling oocytes of each couple were divided equally and inseminated either by conventional ICSI (sperm selected by the embryologist) or inseminated with AI-assisted sperm selection ICSI.
Participants/materials, setting, methods
The study recruited couples with an average female age of 34.7 years, each having at least four mature oocytes at time of ICSI. Sibling oocytes were randomized into two groups: conventional ICSI (n = 134) and AI-assisted ICSI (n = 134). Embryos were cultured in a time-lapse incubator (Embryoscope+), and blastocyst quality was assessed using the AI-driven iDAScore system and ASEBIR grading by an experienced embryologist.
Main results and the role of chance
No statistically significant differences were observed between groups in fertilization rates, blastocyst formation, or embryo quality. However, a trend toward improved blastocyst quality was noted in the AI-assisted group according to ASEBIR grading criteria. The embryo utilization rate was significantly higher in the AI-assisted group (60.4%) compared to the traditional ICSI group (45.5%, p = 0.04). No significant differences were observed in clinical outcomes. These findings suggest that AI-assisted sperm selection provides comparable efficacy to traditional methods while enhancing embryo utilization. The small sample size may limit statistical power, and the findings should be interpreted with caution.
Limitations, reasons for caution
The study’s small sample size and monocentric design limit its generalizability. Larger, multicenter studies are necessary to validate these results and explore the clinical implications further.
Wider implications of the findings
The integration of AI software in ICSI procedures may provide a valuable tool for improving embryo utilization rates and supporting junior embryologists in sperm selection. Broader clinical applications require further investigation.
Trial registration number
No