Artificial intelligence (AI) has emerged as a new medical technology to overcome infertility. By viewing microscopic images and selecting sperm chosen by AI, a 100% in vitro fertilization success rate was achieved. If AI is introduced into infertility treatment, the time and expense required for artificial insemination are expected to be greatly reduced.
Researchers at the Columbia University Center for Infertility said on the 1st that they selected sperm using an AI sperm selection system and succeeded in achieving pregnancy. The study was published the same day in the international journal The Lancet.
Forty percent of infertility is caused by issues on the male side. Ten to 15% of infertile men have azoospermia. Azoospermia means there is no sperm in the semen. In some cases sperm are present, but the number is so small that natural pregnancy, let alone in vitro fertilization, is virtually impossible.
Zev Williams, a professor at the Columbia University Center for Infertility, said, "On the surface the semen appears normal, but under a microscope, in many cases you see only cellular debris instead of sperm," and noted, "Such men are diagnosed as having almost no chance of success not only with natural pregnancy but also with in vitro fertilization."
Patients with azoospermia often undergo surgery to extract sperm directly from the testicles, but this approach has a low success rate and can cause side effects such as inflammation, vascular damage and temporary decreases in male hormones. In addition, technicians sometimes examine semen under a microscope one by one to find the extremely small number of remaining sperm, but this also takes a long time and the expense is very high.
The researchers developed a system called STAR (Sperm Tracking and Recovery) that uses AI to find sperm in the semen of patients with azoospermia. According to lead researcher Professor Hemant Suryawanshi, the system is an integrated platform that combines advanced imaging technology, microfluidics and reproductive endocrinology.
STAR analyzes more than 8 million microscopic semen images per hour. When AI identifies cells with the morphology of sperm, a microchip narrower than a strand of hair isolates the area containing the sperm, and a robotic arm retrieves the sperm. The entire process proceeds in ms (milliseconds, one-thousandth of a second) increments.
The team first applied this technology to a male patient who had tried to conceive for 20 years but failed. They analyzed 3.5 mL of semen provided by the man using STAR and processed 2.5 million images in two hours to find two sperm. These sperm were used for in vitro fertilization with oocytes, creating two embryos (fertilized eggs). They achieved 100% fertilization. One of them successfully implanted in the woman's uterus, resulting in pregnancy.
Professor Williams said, "All it takes for a baby to be born is a single healthy sperm," adding, "This case shows a new possibility for treating male infertility that has long remained unresolved." The researchers said they plan to verify the safety and efficacy of the STAR technology through clinical trials.
References
The Lancet (2025), DOI: https://doi.org/10.1016/S0140-6736(25)01623-X