Collaborative research will help embryologists accurately and quickly identify good sperm
Tokyo, Japan — November 11, 2019 – LPIXEL Inc., Olympus Corporation and the Jikei University School of Medicine made a breakthrough in their joint research by developing an AI-assisted sperm selection system that can calculate sperm motility in real time with high accuracy. The research aims to reduce the intracytoplasmic sperm injection (ICSI) workload and standardize the sperm selection process among embryologists¹ as the demand for assisted reproductive technology grows.
After evaluating 1,066 sperm images in the research, the AI system learned the criteria to recognize sperm in a live image with high accuracy (sensitivity² 99%, positive predictive value³ 92%) and calculate its motility. This innovative AI technology can help embryologists distinguish fast-moving sperm from slow-moving sperm in real time.
Going forward, the research team will develop the AI system to evaluate sperm morphology, including the sperm head and neck. The goals are to further enhance the AI and deploy a microscope equipped with this technology by December 2020. The research team will continue to collaborate to meet these goals.
The AI-powered microscope under development will help to solve challenges in the assisted reproductive technology workflow like the growing workload and lack of standardized selection criteria.
This content was presented by the Jikei University School of Medicine in a paper called, “Study of the Feasibility of Supporting Favorable Sperm Selection by Machine Learning in Microinsemination” at the 64th Annual Meeting of the Japan Society of Reproductive Medicine on November 7 and 8, 2019.
¹A medical technologist who operates assisted reproductive technology, such as ICSI and in vitro fertilization, under the guidance of physicians.
² The rate at which the AI can detect sperm.
³ The percent of real sperm the AI was able to correctly identify.
Example calculation of sperm motility (green: fast, red:slow)
Good sperm (yellow-green) has high straightness and fast. In contrast, low-quality sperm (brown) is poor in straightness and slow. By calculating sperm motility in real time, our AI system can help embryologists quickly identify good sperm in the ICSI workflow.
LPIXEL is a leader in advanced image analysis and processing technology encompassing the life science field. Founded in March 2014, LPIXEL is dedicated to offering research facilities top-tier image analysis technologies and medical diagnosis technologies, both of which adopt advanced AI technology. The company is currently working in collaboration with the University of Tokyo Hospital, the National Cancer Center Japan and other research institutions to further the development of its AI-based medical diagnosis technology. For more information, please visit https://lpixel.net/en/