Presented by Dr. Marcos Meseguer, embryologist at IVI Valencia, at the 34th Congress of the European Society of Human Reproduction and Embryology (ESHRE), celebrated this year in Barcelona.
"This research shows how Artificial Intelligence (AI) is much faster and more consistent than the embryologist to classify embryos using time-lapse images in addition to providing agreement in the processes versus the variability and heterogeneity linked to the work of the human operator ", Explains Dr. Meseguer, author of the study.
This is the conclusion as a result of this work in which the Universidade Estadual Paulista (UNESP) participated, where 5 embryologists from 4 different countries analysed 223 embryos according to the criteria of conventional morphology necessary for embryo selection. The AI has learned to measure, interpret, analyse and distinguish the different parts of the embryo and to select them according to these criteria, perfecting its process as it increases the number of embryos evaluated.
"The application of AI to the classification of the human blastocyst is economical, non-invasive and more reliable than the classification by an operator. Instead of a human looking at thousands of images, the AI continuously evaluates, learns and quantifies additional information. As demonstrated, this technology can inherently improve our ability to assess embryonic viability," he added.
The precise evaluation of embryonic viability and one of the ways to reduce the subjectivity that affects the process of embryo selection, is the use of digital image processing and AI techniques in conjunction with Time-Lapse, which allows us to choose the moment to evaluate the embryo, always at a fixed time, which brings a lot of consistency to the process. This analysis is carried out identically, in any part of the world, since it is performed based on fixed and standardized Time-Lapse images.
In any case, human blastocysts present challenges for the recognition of AI images and therefore further independent studies are required to demonstrate reproducibility before establishing their clinical application.