An Artificial Intelligence-Based Computer Vision Model for Human Sperm Concentration, Motility, and Kinematics...

Area of Science:

• Reproductive Biology
• Medical Technology
• Artificial Intelligence

Background:

• Accurate sperm analysis is crucial for diagnosing and managing male infertility.
• Current methods like manual counting and commercial CASA systems have limitations including labor intensity, human error, and inconsistent performance.

Purpose of the Study:

• To develop and evaluate an AI-driven computer vision tool for high-resolution, quantitative analysis of sperm concentration and motility.
• To benchmark the AI tool against manual tracking and a commercial CASA system.

Main Methods:

• A prospective study analyzed 26 semen samples using an AI computer vision tool, manual tracking (Fiji software), and a commercial CASA system (Hamilton Thorne IVOS II).
• The AI tool computed various sperm motility parameters (VSL, VCL, VAP, LIN, ALHmax, BCF) and concentration.
• Calibration with donor samples ensured accurate concentration mapping.

Main Results:

• The AI tool showed strong linear correlation with manual tracking for VSL, VCL, and VAP (R²=0.93-0.98).
• It outperformed commercial CASA in accuracy and consistency across all motility parameters.
• Post-calibration, RMSE for ALHmax and BCF estimates decreased by 30%-50%.
• AI-based motility grading closely matched manual classification, unlike CASA.
• The AI system demonstrated high repeatability (deviations < ±2%) and robustness under variable imaging conditions.

Conclusions:

• The AI-based tool provides a quantitative, reliable, and repeatable alternative for semen analysis.
• It offers improved accuracy and consistency compared to existing methods.
• This technology supports enhanced fertility diagnostics and potentially more informative treatment strategies.