Deep Learning-Based Morphological Classification of Human Sperm Heads.

Area of Science:

• Reproductive Medicine
• Artificial Intelligence in Healthcare
• Medical Imaging Analysis

Background:

• Male infertility affects over 30% of infertile couples globally.
• Semen analysis, crucial for male fertility assessment, relies heavily on subjective sperm head morphology evaluation.
• Manual sperm morphology assessment is laborious, non-repeatable, and prone to significant inter- and intra-laboratory variability.

Purpose of the Study:

• To develop an automated, accurate, and reliable method for classifying human sperm head morphology.
• To overcome the limitations of manual semen analysis in diagnosing male infertility.

Main Methods:

• A specialized convolutional neural network (CNN) architecture was designed for classifying human sperm head morphology from images.
• The CNN architecture features multiple layers and filter sizes with optimized parameters for efficiency and effectiveness.

Main Results:

• The proposed CNN architecture achieved 88% recall on the SCIAN dataset and 95% recall on the HuSHeM dataset for human sperm head classification.
• The AI model demonstrated superior performance compared to state-of-the-art methods.

Conclusions:

• Deep learning offers a promising solution to enhance the reliability, throughput, and accuracy of sperm morphology assessment.
• The developed AI method has the potential to surpass human embryologists in sperm head classification for male infertility diagnosis.