New AI tool outperforms doctors in accurately counting sperm in infertile men

According to the Cleveland Clinic, 10 percent of all males in the US seeking to conceive are infertile.
Tejasri Gururaj
Representational image
Representational image


Infertility in men is a common issue affecting more than five million couples in the United States (US). According to the Cleveland Clinic, 10 percent of all males in the US seeking to conceive are infertile. 

There are many different ways infertility can manifest in men; non-obstructive azoospermia, or NOA, is one of them. NOA is a condition in which there is no sperm present in the semen. NOA is regarded as the most severe type of infertility, affecting around one percent of all males and five percent of couples seeking fertility treatments.

Currently, men with NOA undergo invasive procedures to retrieve sperm from their testes for use in fertility treatments like Intracytoplasmic Sperm Injection (ICSI).

Now, scientists at the University of Technology Sydney have developed an artificial intelligence (AI) tool called SpermSearch that can quickly and accurately identify sperm in severely infertile men. 

The algorithm was presented at the European Society of Human Reproduction and Embryology. 

ICSI treatment and conventional methods

Conventional treatments for NOA, including testicular sperm aspiration (TESA), testicular sperm extraction (TESE), and microdissection testicular sperm extraction (Micro-TESE), aim to retrieve viable sperm from the testes. 

In the most severe cases, ICSI, a specialized form of in vitro fertilization (IVF), is used. It is commonly used with the retrieved sperm to facilitate fertilization by directly injecting a single sperm into an egg, bypassing natural sperm-egg interaction.

Embryologists have to manually extract the sperm from the samples to fertilize the egg. This process can take up to six hours, affecting the embryologist’s ability to identify sperm because of mental and physical exhaustion.

The new SpermSearch method

The team developed SpermSearch using a systematic methodology. They collected thousands of microscope photos featuring sperm, cells, and other debris. This served as the training data for their AI algorithm. 

The team fed these images as training data to the machine learning algorithm used in SpermSearch. The algorithm learned to recognize and differentiate between sperm and surrounding elements. 

To test the AI tool, the team collected testicular tissue samples from seven patients diagnosed with NOA who had previously undergone surgical sperm removal. The AI algorithm and an experienced embryologist simultaneously examined the sperm samples. The identification performed by the embryologist served as the benchmark for comparison and was considered 100 percent accurate.

The team measured the performance of the AI tool using two metrics: time taken to identify sperm and accuracy. The results showed that the algorithm outperformed the embryologist on both fronts. 

The algorithm was able to identify 611 sperm compared to the embryologist, who only found 560. Moreover, the AI tool demonstrated greater accuracy, finding 60 more sperm and being five percent more precise per viewable droplet area than the embryologist.

The team presented the study as a proof-of-concept, and further clinical trials are required to test the effectiveness and usefulness of SpermSearch. The researchers added that this tool should be tested on men with other forms of infertility, who undergo surgical procedures. 

In a press release, the lead author of the study, Dale Goss, from the University of Technology Sydney, said, "This tool has the ability to give patients who have very little chance of fathering their own biological children an increased chance.

The algorithm improves antiquated approaches that have not been updated in decades. It will ensure the rapid identification of sperm in samples, which will not only increase the chance of a couple conceiving their own biological children but also reduce stress on sperm and increase efficiency in the laboratory."

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board