Aftificial Inteligence in Assisted Reproductive Techniques to Assess Oocyte Quality and Embryo Ploidy

Overview

The assisted reproduction success rate is affected by several factors including the age of the women, oocyte quality and maturation state, as well as sperm quality. Imaging of the meiotic spindle may be crucial for determining the oocyte maturation. Artificial intelligence (AI) will be applied to establish the complex oocyte quality, embryo ploidy and pregnancy success probability from the sequence of data, starting with the recording of the meiotic spindle in polarized light, through paternal factors up to the time lapse recording of early embryo development. This strategy should reduce the cost of fertility treatment thanks to increased efficiency in choosing the most promising candidates and reducing the need for costly laboratory analyses.

One of the main strategies of infertility treatment is in vitro fertilization (IVF). The IVF success rate is affected by several key factors including the age of the women, oocyte quality and maturation state, as well as sperm quality. It has been suggested that the presence, position and retardance of the optically birefringent meiotic spindle (MS) are related to oocyte developmental competence, affecting the quality of fertilization and embryo development. Artificial intelligence (AI) will be applied to establish the complex oocyte quality, embryo ploidy and pregnancy success probability from the sequence of data, starting with the recording of the meiotic spindle in polarized light, through paternal factors up to the time lapse recording of early embryo development. Synergic approaches will be used to increase the quality of embryos for implantation: image analysis and machine learning techniques will be applied to the oocyte microscopic images to perform the MS analysis fully automatically and to determine whether some other aspects of the oocyte appearance might correlate with the optimal timing and fertilization and pregnancy success, or genetic defects. An automatic method of embryo evaluation based on time-lapse videos after ICSI and MS imaging plus other scalar parameters (extracted features can be used as inputs for the downstream tasks, e.g. features extracted from oocytes and sperm can serve as additional inputs to the embryo classifier) will be used. This strategy should reduce the cost of fertility treatment thanks to increased efficiency in choosing the most promising candidates and reducing the need for costly laboratory analyses. The analysis will be performed in cooperation with Czech Technical University and Institute of Physics Academy of Sciences of the Czech Republic who will create a software tool capable of predicting the probability of pregnancy and embryo ploidy status from oocyte images plus time-lapse video of a developing embryo after ICSI. It will be determined whether some other aspects of the oocyte appearance correlate with the fertilization and pregnancy success, or genetic defects. Time lapse sequences of embryonic development and oocyte images will be acquired from VFN and from cooperating IVF centres (Gynem, s.r.o., Repromeda, s.r.o.). The sequences will be stored and paired with outcome (ploidy status, pregnancy) and also with previously acquired oocyte images. BIOCEV (Academy of sciences of the Czech Republic) will evaluate sperm parameters with respect to oocyte fertilization rate and early embryonic development.