An embryo’s return to life: new AI study yields promising results

• A new study, led by researchers at IVI, uses AI to analyse the behaviours of thawed blastocysts to predict implantation potential
• A frozen embryo’s behaviour during the first moments after thawing can help to identify embryos with up to 30% higher chances of achieving pregnancy

Traditionally, embryologists study an embryo’s morphology to gain information about its chances of implantation. In recent years, time-lapse imaging has provided a more precise understanding of embryo development, identifying different morphokinetic parameters as markers of embryo viability and offering complementary models of embryo selection.

However, after freezing and thawing procedures, blastocysts undergo multiple morphological changes that may make it difficult to evaluate their quality, and little is known so far about the application of this technology to vitrified (frozen) and devitrified (thawed) blastocysts.

To explore this further, Dr Marcos Meseguer, IVI Scientific Supervisor and embryologist at IVI Valencia, developed the study titled: “Analysis of the morphological dynamics of blastocysts after vitrification/warming: defining new predictive variables of implantation”, which was presented at the 10th International IVIRMA Congress in Malaga between 20 – 22 April 2023.

“In this study, we evaluate post-thawing dynamics of embryos to predict the implantation potential of embryos that were previously frozen using AI-based Artificial Neural Networks (ANN). In this sense, we are working on an AI algorithm that studies the embryo’s behaviour from devitrification to transfer, which lasts approximately four hours.

AI shows us that an embryo that begins its expansion early (when the average expansion time is 50 minutes) and carries out this process quickly, acquiring a surface area of more than 0.14 square millimetres, is 30% more likely to implant than an embryo that expands later and more slowly during those first four hours of life.

This technology also allows us to identify embryos that, despite showing good morphology, have a low probability of implanting because when they thaw, they have either taken a long time to expand or have expanded very little,” explained Dr Meseguer. 

The retrospective analysis of 511 thawed blastocysts was developed to better understand the embryo re-expansion process following vitrification and subsequent devitrification through the study of morphological dynamics.

“When we vitrify an embryo, we leave it in an inert state by removing water, which is what drives all the machinery in the cell. The moment you remove the water, it is as if time stops, and the embryo can remain like this for years without time having a negative impact on its quality.

When we reactivate time, we add the water back into the embryo. This process is gradual and not all embryos go through this process in the same way. The way the water enters and the ‘antifreeze’ – a cryoprotectant we use to protect the embryo – exits is not the same for all embryos, nor do all embryos start this process at the same time either.

And this is the starting point of our work: we have observed those embryos in which the water begins to enter earlier have a better prognosis. And embryos that expand faster will do better than those that expand more slowly. This leads us to correlate the re-expansion of devitrified blastocysts with their chances of implantation. Thus, more than 60% of the re-expanded blastocysts were successfully implanted, compared to 6% of those that were not re-expanded after thawing,” said Dr Meseguer.

Nowadays, extended culture and blastocyst-stage transfer is common practice, showing improved embryo selection and success rates of IVF treatments. This often involves freezing the remaining viable blastocysts and undergoing subsequent transfers at later dates, minimising the risk of ovarian hyperstimulation (OHSS).

This increasing number of frozen embryo transfer (FET) cycles has improved the precision of embryo selection processes to drive better outcomes for patients.

“It’s well-established that each observation outside the controlled environment of an incubator involves exposure to suboptimal conditions, which can potentially affect treatment success. That’s why continuously monitoring thawed blastocysts using time-lapse systems can provide valuable information about their implantation potential while they remain in a stable, controlled culture environment.

At this point, it is important to highlight that all blastocysts were frozen and thawed using the Cryotop method, and they were placed in the EmbryoScope immediately after devitrification until transfer. Moreover, another differentiating feature of our study is that it provides objective quantitative values for the variables involved in blastocyst re-expansion, as opposed to the subjective morphological evaluation used for blastocyst re-expansion to date,” concluded Dr Meseguer.

Therefore, we can conclude that AI-enabled analysis of thawed embryos could be useful in predicting their implantation potential and using these predictive models in FET cycles could avoid transferring embryos with low chances of implantation. The correlations observed and the proposed algorithm should be validated in a prospective trial to evaluate their efficacy.