Carlos Lobos, business unit director at Pharmaq Analytiq and Eduardo Pulgar, scientific director of Cedai Aquaculture.

Innovative tool with artificial intelligence allows early analysis of embryos

Pharmaq Analytiq introduced a new tool that uses advanced algorithms to diagnose and improve embryonic viability in Atlantic salmon, achieving an accuracy of 85-90%.

Published Last updated

In the auditorium of the Chilean Safety Association in Puerto Montt, Chile, Pharmaq Analytiq held a technical talk with the aim of presenting the results of the project "Diagnosis of Salmo Salar eggs based on morpho-functional embryonic quantification".

The initiative, which was supported by a Crea and Validate I+D+i Business Reactivate fund from Corfo, culminated in the development of a tool that offers an estimated accuracy of 85-90%, providing producers with critical data to make informed decisions and improve efficiency in the production of eggs and embryos of Atlantic salmon.

The event began with remarks from David Espinoza, acting regional director of Corfo Los Lagos and Luis Cárdenas, regional minister of Economy, Promotion and Tourism of the Los Lagos Region, and a talk by Carlos Lobos, business unit director of Pharmaq Analytiq.

Then, Eduardo Pulgar, scientific director of Cedai Aquaculture, biochemist, doctor in biomedical sciences and specialist in reproduction and embryonic development, was in charge of providing more details about the project results. 

The expert began by explaining that the project arose from concerns of the industry, but based on the fact that 50% of the productive mass is lost only in the freshwater stage, clearly influencing the subsequent cultivation cycle in the sea. 

Embryonic Development

With this, the main objective of the project was to understand the causes and moments in which the highest mortalities of Atlantic salmon embryos occur. For this, they analyzed more than 16,000 embryos and 2,000 fish from three fish farms, allowing to generate a broad set of data that were essential for the development of diagnostic tools based on artificial intelligence.

Pulgar explained that the study focused on embryonic development from fertilization to hatching, using advanced deep learning techniques to create algorithms capable of classifying and predicting embryonic viability.

One of the notable achievements was the creation of a classifier that, using images of embryos at different stages of development, can determine the fertilization rate and detect phenotypes with high probabilities of loss.

The analysis revealed that approximately 27% of the oocytes do not fertilize, which could be related to problems with the gametes or their handling. Additionally, it was identified that 34% of the fertilized embryos are lost during the cleaning stage. These findings underscore the importance of optimizing environmental and genetic conditions from the beginning of embryonic development.

“We found an average fertilization rate in fish farms of 73%, meaning there is a 27% that does not get fertilized. This could indicate that the problem lies in the gamete or possibly in handling. This needs to be investigated to understand why this fertilization rate is occurring. And then, in the cleaning stage, 34% of what was fertilized is lost. So, in total, we have losses due to fertilization and losses that occur during the productive stage,” explained the specialist from Cedai Aquaculture.

Then, from the fertilized embryos, the new tool has the ability to detect “abnormal” phenotypes of which about 70% are lost, something that can be identified early on.

Another application shown by Pulgar was the creation of an algorithm to define embryonic viability in the gastrula phase, being able to classify normal, altered, or unfertilized embryos, allowing for more accurate prediction of productive outcomes.

Fry

Finally, the scientist also revealed the development of an algorithm to analyze the heart rate of fry, identifying arrhythmias that can influence their development and survival.

"We created an algorithm that basically studies the heart rate, which on average in fry rearing is about 130 beats per minute when normal. The percentage of fry with arrhythmias can vary but sometimes could be very large, and this correlates a lot with fish that do not follow the productive process or that have a lot of size dispersion," detailed Dr. Pulgar.

With all his results, the expert concluded that on average there is a 35% loss of embryonic production mass during the embryonic stage, and that during the fry stage almost 12% is lost, solely due to morphological problems.