Automated Sperm Head Detection Using Intersecting Cortical Model Optimised by Particle Swarm Optimization.

Area of Science:

• Biomedical Engineering
• Computational Biology
• Medical Imaging

Background:

• Accurate human sperm segmentation is crucial for sperm motility analysis.
• Existing methods face challenges in precise sperm head detection and parameter optimization.
• The Laplacian of Gaussian filter is a common pre-processing step for sperm image segmentation.

Purpose of the Study:

• To propose an optimized intersecting cortical model (ICM) for accurate human sperm head segmentation.
• To enhance the robustness and accuracy of automated sperm segmentation algorithms.
• To improve the efficiency of sperm motility analysis through advanced image processing.

Main Methods:

• Implementation of a Laplacian of Gaussian filter for pre-processing sperm images.
• Development of an intersecting cortical model (ICM) for sperm head segmentation.
• Optimization of the ICM network using particle swarm optimization with feature mutual information as the fitness function.

Main Results:

• The proposed optimized ICM method demonstrated superior accuracy, sensitivity, specificity, and precision compared to four state-of-the-art methods.
• Achieved high performance metrics: 98.14% accuracy, 98.82% sensitivity, 86.46% specificity, and 99.81% precision on 1200 sperm images.
• The optimized algorithm proved to be robust and capable for automated sperm motility analysis.

Conclusions:

• The optimized ICM algorithm offers a significant advancement in automated human sperm segmentation.
• This method provides a robust and accurate tool for clinical applications in sperm motility analysis.
• The algorithm's high performance suggests its potential for widespread implementation in reproductive health diagnostics.