US Applies Neural Network for IVF in Severe Infertility Case

Photo Source: columbiadoctors.org

The IVF procedure was performed at the Center for Reproductive Medicine at Columbia University. The couple described in the publication had been unable to conceive for nearly 19 years. The cause of years of failure was a severe form of male infertility, azoospermia, in which sperm are absent or found in single quantities. According to research data, azoospermia occurs in 10-15% of men with infertility, accounting for approximately 4-6% of all infertile couples.

To solve the problem, scientists used the AI system STAR (Sperm Tracking And Recovery). How the technology works: a high-speed camera takes approximately 8 million images of the sample per hour, then the neural network identifies sperm among cellular debris in real time. A microfluidic chip with channels as thick as a human hair isolates the portion of the sample with the found cell, after which a robot extracts the sperm in milliseconds for subsequent use in IVF.

In the described case, the system found 2 viable sperm from a 3.5 ml sample, which were used to create two embryos. One embryo was transferred, and pregnancy occurred.

The authors note that this is the first documented case of successful pregnancy using sperm found and extracted by an AI-controlled microfluidic platform.

The technology should significantly accelerate diagnosis and treatment of severe forms of male infertility. During system testing, in one sample embryologists manually searched for sperm for 2 days and found none, while STAR detected 44 sperm in 1 hour.

AI is primarily used in IVF for monitoring development and selecting embryos after artificial fertilization. In Russia, this technology is applied, for example, in the embryology laboratory of the E.G. Lebedeva Center for Innovative Reproductive Technologies at Mother and Child Clinical Hospital. Alexander Vysotsky, laboratory director, discussed the process specifics in an interview with Marus Media.

Source: The Lancet