Researchers from the University of the Basque Country (UPV-EHU) have developed a new mathematical classification scheme that can be used to select embryos for use in assisted reproduction treatments. The Spanish scientists presented this 'intelligent system' of support for infertility treatments in the journal Computer Methods and Programs in Biomedicine.
'Up to now experts working in IVF have selected the best embryos subjectively, based on their training and experience', said Dinora A. Morales, from the Intelligent Systems Group at UPV-EHU. Instead, the new mathematical classifiers provide a formal framework to allow embryologists to identify the healthiest embryos that are most suitable for use in IVF treatment. Spanish law allows the transfer of up to three embryos at once to a woman's uterus, and so the researchers monitored the evolution of trios of embryos in 63 cases from the infertility programme at Clinica del Pilar in San Sebastian (Guipuzcoa), and used this information to help develop the scheme.
First of all the case history of each infertile couple was noted (including age, type of infertility and sperm quality) as well as the form and structure of the zygotes (fertilised eggs) and the resulting embryos. Then, from microscope images, the scientists managed to measure and classify the zygotes and embryos, the blastomeres (cells produced by the division of the zygote) and their degree of fragmentation, and the thickness of their surrounding membrane known as the 'zona pellucida'. Bayesian classifiers (an application of Bayes' probability theorem) were used to process this information and calculate the probability of successful implantation resulting from an embryo being transferred to a woman's uterus. 'These types of mathematical classifiers provide experts with evidence on what embryo characteristics enable the identification of the most ideal embryos, through the selection of variables,' Morales explained.
The Spanish researchers also published the results of a follow-up study, in the journal Computers in Biology and Medicine, where they compared the effectiveness of different Bayesian classifiers as a tool for choosing the best embryos. By analysing 249 photographs of embryos from the database held at the Genesis Centre in Rome (Italy) they discovered that the 'wrapper-TAN' classifier had a success rate of over 90 per cent. Morales and her team plan to continue their research by collaborating with other hospitals to perfect these techniques for selecting the best candidate embryos for use in infertility treatment. In particular they would like to focus on predicting multiple pregnancies which are associated with greater risks for both women and babies.
Sources and References
Mathematicians make better babies