Major breakthrough for male fertility diagnosis driven by AI

The process of receiving a fertility diagnosis can be slow and challenging, with one in seven couples struggling to conceive, according to the NHS. The National Library of Medicine states that 30–50% of the time male infertility contributes to this, an issue too often overlooked, despite impacting an estimated 7% of men. The ever decreasing rate of male fertility has seen sperm counts drop by more than 50% over the past 40 years, prompting the team at Bayezian to approach it from a data science standpoint, with which they have  been able to identify sperm fertility with 96% accuracy rate – 2% higher than existing scientific approaches.

Bayezian began the project 18 months ago after being approached by a concerned group to improve the way fertility is diagnosed. The team used ‘The Modified Human Sperm Morphology Analysis (MHSMA)’ dataset to build deep learning frameworks that can see a sperm’s morphology. With differences that the human eye can’t see, the Bayezian team has built an algorithm that can determine how healthy a sperm is, and ultimately whether that sperm is fertile.

The MHSMA dataset includes a collection of human sperm images from over 200 patients with male factor infertility. Each image is labelled by experts for normal or abnormal sperm acrosome, head, vacuole, and tail. This dataset can then be used to conduct Sperm Morphology Analysis by conducting deep learning methodologies.

The project was part of Bayezian’s Incubator, which combines academic excellence, industry expertise, and a commitment to advancing science to solve real-world problems.

Ed Dixon, Founder and CEO at Bayezian, commented: “By approaching real world challenges from a data science perspective, new capabilities can be achieved. This project is the perfect example of the tech for good approach that the team is undertaking. We see accurate diagnosis as a critical tool in helping address male fertility.”