Web- and Artificial Intelligence–Based Image Recognition For Sperm Motility Analysis: Verification Study (Preprint)

Abstract

BACKGROUND Human sperm quality fluctuates with times. Therefore, it is crucial for couples preparing for natural pregnancy to monitor sperm motility. CreateCare corporate developed a smartphone-based sperm motility test (Bemaner ®) composed of microscope and microfluidic modules and designed to adapt to different types of smartphones. OBJECTIVE This study is to verify the performance of AI image recognition and cloud computing of the system. METHODS Sperm videos are captured and uploaded to the cloud by smartphones software. Analysis of sperm motility is performed by AI image recognition algorithm with cloud computing then results are displayed for users. According to the number of motile sperm in vision field, forty-seven delinked sperm videos are scored into 6 grades (0-5) by a male-fertility expert of 10-year experience. Pearson’s product-moment correlation is calculated by Excel (Microsoft® 2013) between the scored grades and results (concentration of total sperm, concentration of motile sperm and motility percentage) computed by the system. RESULTS Good correlations are demonstrated between the scored grades and results computed by the system with AI algorithm (v 1.0.8_5/22). The respective correlation coefficients are 0.65, 0.84, and 0.90, and all P values are below .001. CONCLUSIONS This smartphone-based sperm motility test (Bemaner®) accurately measures motility-related parameters and could potentially be applied towards the following fields: male infertility detection, sperm quality test during preparation for pregnancy, and infertility treatment monitoring. With frequent at-home testing, more data can be collected to help make clinical decisions and conduct epidemiological researches. CLINICALTRIAL None. Because this study was to compare examining results between by human and by AI on sperm videos from an archive, no patients/subjects' personal data and parameters were involved.