Artificial intelligence and hyperspectral imaging: a new approach to optimizing sperm selection

Advances in reproductive medicine have increased the likelihood of pregnancy for millions of people worldwide. However, there are still aspects of the reproductive process that continue to be researched and improved. One of these is the selection of the sperm to be used during assisted reproduction treatments.

The integration of tools based on hyperspectral imaging techniques and artificial intelligence is opening up new possibilities in this field. These technologies make it possible to study sperm characteristics that, until now, could not be analyzed without compromising their integrity, providing information that could be linked to reproductive success.

The results of this research, presented by IVI RMA Global and the IVI Foundation, were unveiled at the 42nd Annual Meeting of the European Society of Human Reproduction and Embryology (ESHRE).

Beyond morphology and motility

Sperm selection has traditionally been based on assessing visible characteristics; primarily, their shape and movement. These variables represent only a portion of the cell’s biological information, however.

In procedures like in vitro fertilization (IVF), and ICSI, all available oocytes are used during treatment, while a large number of sperm are discarded.

As Dr. Nicolás Garrido, director of the IVI Foundation and principal investigator of the study, explains, “In a treatment, we use all the oocytes, but we discard millions of sperm. And each one is genetically distinct and can lead to a different outcome.”

This underscores the need to deepen our understanding of the male gamete and how its internal characteristics can influence embryonic development.

What hyperspectral imaging offers

Hyperspectral imaging provides biochemical information about cells by analyzing how light interacts with tissues.

Unlike conventional observation techniques, this method records signals across multiple wavelengths, yielding data on the molecular composition of the sperm.

“Hyper-spectral imaging is a game-changer, as it lets us gather biochemical information from the sperm without damaging it. We can later use the very same sperm cell that we analyzed,” says Dr. Garrido.

The ability to study the internal structure of sperm without affecting its viability is one of this technology’s major breakthroughs.

Artificial intelligence to interpret complex data

The information generated by hyperspectral imaging is extensive and requires tools capable of processing large volumes of data, which is why Artificial Intelligence is necessary.

In this context, artificial intelligence makes it possible to identify patterns and associations that would be difficult to detect using traditional methods.

“AI lets us turn previously inaccessible information into analyzable data. We’ve gone from observing shape and movement to being able to analyze and understand the role played by the molecular content of the sperm,” explains Dr. Garrido.

Thanks to these algorithms, researchers can study the relationship between specific molecular characteristics and a sperm cell’s ability to produce a viable embryo.

Predicting reproductive potential

One of the goals of this study is to develop tools that can estimate the likelihood of success for each sperm cell.

“Hyper-spectral imaging analysis lets us design statistical models to predict the individual probability of each sperm becoming a viable blastocyst,” explains Dr. Garrido.

The use of predictive models represents a step forward from traditional selection criteria, as it incorporates additional biological information that can aid in decision-making and success.

Impact on assisted reproduction treatments

More precise sperm selection could directly impact treatment effectiveness, which is always the ultimate goal of innovation.

“If we can select the sperm with the greatest potential, we enable patients to have children sooner, with a higher chance of success and less need for repeat treatments,” explains Dr. Garrido.

Potential benefits for patients include:

• Fewer cycles required.
• Reduced emotional burden associated with the treatment.
• Enhanced clinical efficiency.
• Increased chances of success.

Moving toward more personalized reproductive models

Current research aims to integrate various sources of information to better understand the interaction between gametes.

“We are developing models that combine hyper-spectral imaging of the sperm with that of the oocyte, as well as the patient’s clinical characteristics,” explains Dr. Garrido.

This approach recognizes that embryonic development depends on both the sperm and the oocyte.

“The embryo doesn’t depend solely on the sperm. The oocyte also plays a role, so understanding that interaction will be key to further improving results,” he adds.

Incorporating clinical and biological variables could help develop more personalized treatments tailored to each patient.

Next steps for clinical application

Although the results are promising, this technology still needs to be fully validated before it can be incorporated into routine clinical practice.

“Our plan is to complete these studies in about a year, and then work on integrating them into clinical practice,” concludes Dr. Garrido.

The combination of artificial intelligence and hyperspectral imaging marks a new era in assisted reproduction, moving toward more precise, individualized, and evidence-based reproductive medicine.

The ability to access previously unseen information can help improve sperm selection and, as a result, increase the chances of a treatment’s success.

Dr. Nicolás Garrido

Director of the IVI Foundation and Head of Research Administration at IVI RMA Global. Specialized in male infertility and member of the Scientific Committee at FIVI Valencia.

Find out more