Prediction of embryo implantation by machine learning based on ultrasound strain imaging

Because of the trend to postpone childbirth, the rate of couples dealing with infertility is rapidly increasing and approaching 20%. In-vitro fertilization (IVF) represents the only reproduction option in Europe for 2.5 million couples. However, its success rate remains below 30%. There is clear consensus on a major involvement of uterine contractions in IVF failure, especially during and after embryo transfer. Quantitative and non-invasive measurement of uterine (peristaltic) activity, combined with accurate interpretation and classification methods, can provide an important contribution towards improved IVF success rates. Therefore, this study investigates the use of machine learning for probabilistic classification of the uterine activity, as either favorable or adverse to embryo implantation. The results obtained in 16 patients undergoing an IVF cycle confirm the ability to predict successful embryo implantation by ultrasound uterine motion analysis combined with machine learning.